Upload
lamdat
View
236
Download
0
Embed Size (px)
Citation preview
Thesis submitted for the degree of
Masters of Information Technology (Research, IT60)
Centre for IT Innovation
Faculty of IT
Queensland University of Technology
Australia
Reference Models for IT Service Provision
Author: Chris Taylor
Supervisor: Associate Professor Dr Michael Rosemann
Associate Supervisor: Associate Professor Dr Glenn Stewart
Year of Submission: 2003
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page i
Key words Reference Model
IT Service Provision
ITIL
Quality of Models
Business Process Management
Knowledge Management
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page ii
Abstract The new age, the “Information Age” (Davenport and Prusak 1997) has
increased society’s and businesses’ reliance on Information Technology (IT).
Hence, there is an increasing focus on the management of IT, not only from a
technological perspective but also from a business perspective.
This research, sponsored by REALTECH and the Australian Research
Council, applies one of the modern management approaches, business
process management (Hammer 1990), to the domain IT service provision, by
designing a business process reference model for IT Service Provision.
A reference model is an abstracted depiction of reality that serves as a
standardised or suggestive conceptual basis for the design of enterprise
specific models, usually within a like domain. They are one method of
improving the efficiency and effectiveness of enterprise modelling and can
also be used to standardise communication or capture knowledge.
There is a general lack of theory regarding the classification, design and
quality of reference models. The first part of this thesis attempts to fill these
gaps, by presenting a reference model classification scheme, outlining 7
philosophies for the design of reference models and detailing 2 case studies
on the user-perceived quality of business process reference models.
Reference models and the Business Process Management Lifecycle
(Rosemann 2000) are integrated to show how reference models can be
applied to improve the efficiency and effectiveness of business process
improvement projects.
This reference model theory was then applied to produce a model for domain
of IT Service Provision. Investment in IT has increased to become the largest
single element of capital expenditure (Thorp 1998). Gartner predicted that
organisations will spend over 10% of revenue on IT by 2005 (Haines 2000).
A major input for this model is the ITIL best practice documents (CCTA
2000). The reference model focuses on Incident Management and used
focus groups with participants from several large IT service providers to
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page iii
validate the models. The designed reference model is then tested in two case
studies to determine its accuracy and usefulness.
The thesis finishes with a discussion of the designed model, the
effectiveness of the procedural model and provides suggestions for the
design of other reference models. The final chapter provides a summary and
an outlook for further research into the area.
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page iv
Table of Contents
KEY WORDS ......................................................................................................................................... I
ABSTRACT...........................................................................................................................................II
TABLE OF CONTENTS..................................................................................................................... IV
LIST OF FIGURES............................................................................................................................VIII
LIST OF TABLES ............................................................................................................................... XI
STATEMENT OF AUTHORSHIP.....................................................................................................XII
ACKNOWLEDGEMENTS ...............................................................................................................XIII
CHAPTER 1: PROBLEM OUTLINE ......................................................................... 1 1.1 Introduction.................................................................................................................... 1 1.2 REALTECH Australia .................................................................................................... 3 1.3 Research Objective and Research Questions ................................................................. 3 1.4 Thesis Structure.............................................................................................................. 4
CHAPTER 2: LITERATURE REVIEW...................................................................... 6 2.1 Introduction.................................................................................................................... 6 2.2 Reference Models ........................................................................................................... 6
2.2.1 Definition ................................................................................................................................. 6 2.2.2 Theoretical Justification ......................................................................................................... 12 2.2.3 Reference Model Lifecycle .................................................................................................... 13
2.3 Classification of Reference Models .............................................................................. 14 2.3.1 Characteristics from Literature............................................................................................... 15 2.3.2 Derived Classification Characteristics ................................................................................... 21 2.3.3 Characteristics from a Review of Reference Models ............................................................. 31 2.3.4 Classification Summary ......................................................................................................... 33 2.3.5 Example of Reference Model Classifications ........................................................................ 35
2.4 Typical Applications of Reference Models ................................................................... 37 2.5 Business Process Modelling ......................................................................................... 39 2.6 Applying Business Process Reference Models ............................................................. 42
2.6.1 Templates............................................................................................................................... 46 2.6.2 Scope Definition and Targeting Suggestion........................................................................... 54 2.6.3 Process Benchmark................................................................................................................ 55 2.6.4 Implementation Guide............................................................................................................ 55 2.6.5 Run Time Information ........................................................................................................... 56 2.6.6 Performance Benchmark and Template ................................................................................. 56
2.7 IT Service Provision ..................................................................................................... 56 2.7.1 Application Service Providers................................................................................................ 58 2.7.2 Taxonomy of IT Services....................................................................................................... 61 2.7.3 IT Services Lifecycle ............................................................................................................. 63
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page v
2.8 ITSP Business Reference Models ................................................................................. 65 2.8.1 Desired IT Service Provider BP Reference Model................................................................. 66 2.8.2 Existing Models related to IT Service Provision.................................................................... 67 2.8.3 Perceived Shortfalls in existing ITSP BPRM......................................................................... 81
2.9 Chapter Summary......................................................................................................... 82 CHAPTER 3: RESEARCH DESIGN AND METHOD ................................................ 84
3.1 Introduction.................................................................................................................. 84 3.2 Research Design........................................................................................................... 84 3.3 Focus Group................................................................................................................. 86
3.3.1 Introduction............................................................................................................................ 86 3.3.2 Characteristics........................................................................................................................ 86 3.3.3 Conducting Focus Groups...................................................................................................... 86 3.3.4 Justification............................................................................................................................ 87
3.4 Case Study .................................................................................................................... 88 3.4.1 Introduction............................................................................................................................ 88 3.4.2 Characteristics........................................................................................................................ 88 3.4.3 Conducting Case Studies........................................................................................................ 89 3.4.4 Justification............................................................................................................................ 90
3.5 Chapter Summary......................................................................................................... 90 CHAPTER 4: REFERENCE MODEL QUALITY ..................................................... 91
4.1 Introduction.................................................................................................................. 91 4.1.1 Motivation.............................................................................................................................. 92
4.2 Terminology ................................................................................................................. 92 4.3 Research Question........................................................................................................ 93 4.4 Research Design........................................................................................................... 94 4.5 Quality Framework ...................................................................................................... 95
4.5.1 Quality of Modelling Framework .......................................................................................... 95 4.5.2 Guidelines of Modelling ........................................................................................................ 96 4.5.3 Semiotic Framework .............................................................................................................. 96 4.5.4 Deriving the Quality Framework ........................................................................................... 96
4.6 Case Study Descriptions............................................................................................. 100 4.7 Data Collection .......................................................................................................... 101 4.8 Data Analysis ............................................................................................................. 102 4.9 Findings...................................................................................................................... 102 4.10 Limitations and Conclusions ...................................................................................... 107 4.11 Chapter Summary....................................................................................................... 109
CHAPTER 5: REFERENCE MODEL DESIGN PROCEDURE ................................. 110 5.1 Introduction................................................................................................................ 110 5.2 Reference Model Design Philosophies ....................................................................... 111
5.2.1 Blue Sky Design .................................................................................................................. 112 5.2.2 Design by Choice ................................................................................................................. 113 5.2.3 Baseline Design ................................................................................................................... 113
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page vi
5.2.4 Common Practice Design..................................................................................................... 114 5.2.5 Best Practice by Composition .............................................................................................. 115 5.2.6 Design by Abstraction.......................................................................................................... 116 5.2.7 Explicit Alternatives Design ................................................................................................ 118 5.2.8 Design Philosophy Summary............................................................................................... 119
5.3 Reference Model Design Procedural Model .............................................................. 120 5.3.1 Define the Reference Model ................................................................................................ 122 5.3.2 Design .................................................................................................................................. 123 5.3.3 Validate................................................................................................................................ 126 5.3.4 Test ...................................................................................................................................... 126 5.3.5 Select Low-Level ................................................................................................................. 127
5.4 Chapter Summary....................................................................................................... 127 CHAPTER 6: ITSP REFERENCE MODEL DESIGN............................................. 128
6.1 Introduction................................................................................................................ 128 6.2 Define the Reference Model ....................................................................................... 128
6.2.1 Model Purpose ..................................................................................................................... 128 6.2.2 Model Characteristics .......................................................................................................... 129 6.2.3 Modelling Conventions........................................................................................................ 135
6.3 Design and Validate High-Level Process Framework ............................................... 145 6.3.1 Design Philosophy ............................................................................................................... 145 6.3.2 Identify Information Entities................................................................................................ 146 6.3.3 Build .................................................................................................................................... 149 6.3.4 Validate................................................................................................................................ 156
6.4 Select Low-Level Process........................................................................................... 159 6.5 Design and Validate the Low-Level Model ................................................................ 159
CHAPTER 7: REFERENCE MODEL TESTING..................................................... 163 7.1 Introduction................................................................................................................ 163 7.2 Test High-Level .......................................................................................................... 163
7.2.1 Case Study Design ............................................................................................................... 166 7.2.2 Case Study Results............................................................................................................... 167 7.2.3 Limitations and Conclusions................................................................................................ 170 7.2.4 Effect on Reference Model .................................................................................................. 172
7.3 Test Low-Level .......................................................................................................... 173 7.3.1 Case Study Design ............................................................................................................... 174 7.3.2 Case Study Results............................................................................................................... 180 7.3.3 Limitations and Conclusions................................................................................................ 185
7.4 Chapter Summary....................................................................................................... 188 CHAPTER 8: ITSP REFERENCE MODEL OUTPUTS.......................................... 189
8.1 Introduction................................................................................................................ 189 8.2 Model Outputs ............................................................................................................ 189
8.2.1 Level 0 Model – Business Process Framework.................................................................... 189 8.2.2 Level 1 Model: Incident Management – VAC ..................................................................... 195
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page vii
8.2.3 Level 2 Models .................................................................................................................... 196 8.2.4 Model Output Summary....................................................................................................... 201
8.3 Methodological Outputs ............................................................................................. 203 8.3.1 Use of Focus Groups............................................................................................................ 203 8.3.2 Bottom-up versus Top-down................................................................................................ 205 8.3.3 Modelling of Process Alternatives ....................................................................................... 206
CHAPTER 9: SUMMARY AND OUTLOOK........................................................... 207 9.1 Thesis Summary.......................................................................................................... 207 9.2 Future Work ............................................................................................................... 213
REFERENCES................................................................................................................................... 217
APPENDICES ................................................................................................................................... 224
Appendix A Quality Case Study Protocol............................................................................. 224 Appendix B Examples of classifications of Levels of Reuse ................................................. 231 Appendix C Survey Forms .................................................................................................... 234 Appendix D Detailed Survey Results .................................................................................... 236 Appendix E Company Profiles ............................................................................................. 237
Citec..................................................................................................................................................... 237 Computer Science Corporation............................................................................................................ 237 Corporate Service Agency ................................................................................................................... 237 Deloitte Touche Tohmatsu................................................................................................................... 238 EDS Consulting ................................................................................................................................... 238 Hitachi Data Systems........................................................................................................................... 239 IBM Global Services Australia ............................................................................................................ 239 Mincom................................................................................................................................................ 239 Parmalat Pty Ltd .................................................................................................................................. 240 Queensland Rail................................................................................................................................... 240 REALTECH AG.................................................................................................................................. 240 Telstra .................................................................................................................................................. 241
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page viii
List of Figures Figure 1: Reference Model Lifecycle and Characteristics_________________________________ 14 Figure 2: Chapter 2.3 in relation to the Reference Model Lifecycle _________________________ 15 Figure 3: Chapter 2.6 in relation to the Reference Model Lifecycle _________________________ 42 Figure 4: Business Process Lifecycle ________________________________________________ 44 Figure 5: Business Process Lifecycle and related use of Reference Model ____________________ 45 Figure 6: Types of re-use and parts of the model re-used _________________________________ 46 Figure 7: Using a Reference Model as a True Template __________________________________ 47 Figure 8: Using a Reference Model as a Semantic Foundation ____________________________ 47 Figure 9: Derivation by Customisation _______________________________________________ 48 Figure 10: Derivation by Red-lining _________________________________________________ 49 Figure 11: Derivation by Configuration ______________________________________________ 49 Figure 12: Derivation by Embellishment______________________________________________ 50 Figure 13: Applications as Template _________________________________________________ 52 Figure 14: Scope and Definition and Targeting Suggestions ______________________________ 54 Figure 15: IT services as defined by Benson (2002) _____________________________________ 58 Figure 16: xSP Taxonomy (from Seymore and Edwards 2001)_____________________________ 60 Figure 17: Four layered stratification of the ASP market (from Tao 2001) ___________________ 60 Figure 18: Simplified Taxonomy of IT Services _________________________________________ 61 Figure 19: IT Services Lifecycle ____________________________________________________ 64 Figure 20: EMS’s High level ITIL process model _______________________________________ 69 Figure 21: Examples of inconsistencies in modelling in ITIL documents _____________________ 70 Figure 22: IT Service CMM Methodology _____________________________________________ 71 Figure 23: The IT Service CMM ____________________________________________________ 72 Figure 24: Microsoft Operating Framework ___________________________________________ 74 Figure 25: Garschhammer et al. Service Model ________________________________________ 76 Figure 26: eTOM Business Process Framework - Level 0 Processes ________________________ 78 Figure 27: Corbit Overview________________________________________________________ 80 Figure 28: Research Design and embedded Methodologies _______________________________ 85 Figure 29: Chapter 4 in relation to the Reference Model Lifecycle__________________________ 91 Figure 30: Quality Terminology ____________________________________________________ 92 Figure 31: Relationships between view, purpose, perspective and quality ____________________ 93 Figure 32: RM Lifecycle phase in which the Quality research is based ______________________ 94 Figure 33: Quality Framework for this paper __________________________________________ 98 Figure 34: RM Lifecycle related to the types of economic efficiency________________________ 107 Figure 35: Conclusions that add to or differ from current literature _______________________ 108 Figure 36: Chapter 5 in relation to the Reference Model Lifecycle_________________________ 110 Figure 37: Examples of possible information entities ___________________________________ 112 Figure 38: Baseline Design _______________________________________________________ 114 Figure 39: Common Practice Design _______________________________________________ 115
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page ix
Figure 40: Best Practice by Composition ____________________________________________ 116 Figure 41: Design by Abstraction __________________________________________________ 117 Figure 42: Relationship between Benefit vs. level of abstraction __________________________ 118 Figure 43: Explicit Alternatives Design______________________________________________ 119 Figure 44: Proposed BP Reference Model Design Procedural Model ______________________ 121 Figure 45: Depiction of the design step of Relating Information Entities ____________________ 125 Figure 46: Purpose of the developed ITSP Reference Model _____________________________ 129 Figure 47: Interaction of Levels of Abstraction________________________________________ 132 Figure 48: Scope of the ITSP Reference Model ________________________________________ 136 Figure 49: Example of error/non-error process split depictions ___________________________ 139 Figure 50: Example showing removal of trivial event ___________________________________ 140 Figure 51: Combining Mandatory and Optional process paths ___________________________ 141 Figure 52: Example of layout of non sequential VAC steps_______________________________ 143 Figure 53: Dimensions in the ITSP Level 0 view _______________________________________ 144 Figure 54: Model Hierarchy Structure ______________________________________________ 145 Figure 55: Combining Blue Sky and Best Practice Design Philosophies ____________________ 146 Figure 56: Porter's Value Chain ___________________________________________________ 151 Figure 57: Highest level of Porter's Value Chain ______________________________________ 151 Figure 58: ITSP first version ______________________________________________________ 152 Figure 59: eTOM (v2.5) Level 1 processes ___________________________________________ 154 Figure 60: ITSP Second Version ___________________________________________________ 155 Figure 61: Draft of the ITSP Model as presented to the focus group _______________________ 156 Figure 62: Reference Model showing "Customer" object on right _________________________ 159 Figure 63: Example of ITIL text and diagrams ________________________________________ 160 Figure 64: Rosemann’s BPM Lifecycle relationship to Huxley's Target methodology __________ 164 Figure 65: High Level model used for first case study (Core processes only)_________________ 165 Figure 66: Enterprise specific model of first case study _________________________________ 167 Figure 67: Enterprise Specific Model for Case study 2 __________________________________ 169 Figure 68: Level 0 Processes______________________________________________________ 173 Figure 69: Process Groupings of the ITSP Reference Model _____________________________ 190 Figure 70: Level 0 Processes of the ITSP Reference Model ______________________________ 191 Figure 71: Level 2 Value Added Chain for Incident Management _________________________ 195 Figure 72: Business Process Lifecycle and related use of Reference Model __________________ 208 Figure 73: Quality Conclusions that add to or differ from literature _______________________ 209 Figure 74: Proposed BP Reference Model Design Procedural Model ______________________ 211 Figure 75: Level 0 Processes of the ITSP Reference Model ______________________________ 212 Figure 76: Example of possible Interaction Model _____________________________________ 216 Figure 77: Example of High Level of Re-Use Classification (Incident Management)___________ 231 Figure 78: Enterprise model example for medium re-use (Detect Incident) __________________ 231 Figure 79: Reference model example for medium re-use (Detect Incident)___________________ 232
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page x
Figure 80: Enterprise model example for low re-use (Record Incident) _____________________ 232 Figure 81: Reference model example of Low Level of Re-Use (Record Incident) ______________ 233
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page xi
List of Tables Table 1: Fettke and Loos (1999) Reference Model Classification Scheme.......................................... 17 Table 2: Existing and Proposed Model Characteristic Type Overlaps................................................ 19 Table 3: Classification of Reference Models........................................................................................ 34 Table 4: Example reference model classification of SCOR.................................................................. 36 Table 5: Examples of uses for Business Process Modelling ................................................................ 42 Table 6: Readiness for Use Type and Type of Derivation Required .................................................... 51 Table 7: Traditional Outsourcing vs. ASP (Yaho and Murphy 2001).................................................. 60 Table 8: Desired characteristics for IT Service Provider BP Reference Model .................................. 67 Table 9: ITIL Classification ................................................................................................................. 70 Table 10: IT Service CMM Classification............................................................................................ 73 Table 11: IT Service CMM Classification............................................................................................ 75 Table 12: Garschhammer et al. Classification..................................................................................... 76 Table 13: eTOM v2.5 Classification .................................................................................................... 79 Table 14: CORBIT classification ......................................................................................................... 81 Table 15: Overlap in the Modelling frameworks and proposed quality dimensions............................ 97 Table 16: Citations of Quality attributes from case study interviews ................................................ 103 Table 17: Relationship between Design Philosophy and required Design Steps............................... 123 Table 18: Summary of ITSP Reference Model Characteristics.......................................................... 135 Table 19: Object Types for eEPCs ..................................................................................................... 138 Table 20: Maintained attributes in eEPCs......................................................................................... 139 Table 21: eEPC connection types ...................................................................................................... 139 Table 22: Object Types for VAC ........................................................................................................ 143 Table 23: Connection Types for VAC................................................................................................. 143 Table 24: Maintained Attributes for VAC .......................................................................................... 143 Table 25: Summary of Differences to Reference Model..................................................................... 170 Table 26: Re-Use of models in the CITEC case study........................................................................ 180 Table 27: Summary of Improvements derived from CITEC Models .................................................. 183 Table 28: Re-Use of models in the Pauls case study.......................................................................... 183 Table 29: Summary of Improvements derived from Pauls Models..................................................... 184 Table 30: Sample Results from Survey questions............................................................................... 185 Table 31: Characteristics of an IT Service Provision BP Reference Model ...................................... 202 Table 32: Reference Model Classification Characteristics................................................................ 207 Table 33: Results of the Likert questions comparing the books and models ...................................... 236
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page xii
Statement of Authorship The work contained in this thesis has not been previously submitted for a
degree or diploma at any other higher education institution. To the best of my
knowledge and belief, the thesis contains no material previously published or
written by another person except where due reference is made.
_________________
Signature of Author
_________________
Date
Masters Thesis Chris Taylor
QUT – Centre for IT Innovation Page xiii
Acknowledgements Several individuals deserve credit for the work contained in this thesis.
Firstly, and principally, the author would like to thank Dr Rosemann for his
vision, and support throughout the 18 months of the project. The author is
greatly appreciative for the opportunity and guidance provided by Dr
Rosemann, and feels privileged to have worked so closely with such a
person.
Secondly, the work would not have started without the significant contribution
of the industry partner REALTECH and the Australian Research Council. A
special thanks to Wayne Baker and Helena Mendes from REALTECH for
their support throughout.
The author is also very appreciative to Mr Grant Erickson from Telstra who
agreed to be part of the internal review panel.
This work relied heavily on the organisations that provided personnel,
valuable time and expertise to contribute through the various phases of the
research. A special thanks to those organisations (listed in the appendix).
Also thanks to Craig Huxley and Dr Christian Probst for constant feedback,
Dr Glenn Stewart as associate supervisor, Wasana Sedera for the
collaboration on the paper submitted to ACIS 2003, Marit Schallert for
support in the student projects and the rest of the CITI staff and students for
making my time easy and thoroughly enjoyable.
I would also like to thank my family who have supported me throughout this
process, as they have throughout the rest of my life.
Masters Thesis Chris Taylor
QUT Page 1
Chapter 1: Problem Outline
This chapter introduces the problem domain and draws from it a problem
definition. It concludes with a description of the thesis structure.
1.1 Introduction
Recent decades have seen a dramatic shift in most economies away from
the “Industrial Age”. The new age, the “Information Age” (Davenport and
Prusak 1997), involves complex business interactions and activities and has
demanded a trend away from the purely function management approaches
that were popularised by authors such as Frederick Taylor (Taylor 1911). As
we move to the “Knowledge Society”, Taylorism as a management tool has
been replaced, or at least supplemented, with trends such as Knowledge
and/or Information Management (Drucker 1990; Hammer 1990; Davenport
and Prusak 1997).
One method of managing this knowledge or information and the increasing
complexity of the Information Age is to codify “knowledge into [an] abstract
system of symbols” (Bell 1973 p20).
The Oxford English Dictionary Online defines the word “model” as:
“a conceptual or mental representation”
Essentially, an “abstract system of symbols” is a model.
Models have been used ever since man first drew on a cave wall, but
recently, driven by the need to support the previously mentioned changes in
society, have seen increased importance. Starting in the 1970s, the
integration of Information Technology into business and society has
increased the need for Enterprise Models i.e. “all the various designs, models
prepared for analysis, executable models to support enterprise operation”
(Bernus et al. 1998 p18). Enterprise modelling, which is the process of
creating these enterprise models, can be a prohibitively expensive activity.
This is particularly true when an enterprise is starting from scratch (Hars
1993). Fortunately, many organisations and individuals have recognised that
often models can be re-used e.g. (Hay 1996; Bernus 1998). The term used to
Masters Thesis Chris Taylor
QUT Page 2
describe a model that can be, and usually has been designed for, re-used in
several situations is “reference model”. These reference models can
significantly reduce the amount of work required for enterprise modelling
(Schütte 1998). Estimates for the re-use of data models include Silverston,
who predicts that around two-thirds of any particular data model could be re-
used in another situation within a particular industry (Silverston et al. 1997).
Even across industry boundaries there are traditional highly generic activities
such as accounting or inventory, allowing a wide range of varied businesses
to benefit from a relatively few number of reference models (Fowler 1997
p10). One of these generic activities that is been becoming increasing
important to enterprises is Information Technology. Willcocks, Feeny and
Islei argue that effective management of IT resources is paramount to
organisational success (Willcocks et al. 1997).
Investment in IT has increased to become the largest single element of
capital expenditure (Thorp 1998). Gartner predicted that organisations will
spend over 10% of revenue on IT by 2005 (Haines 2000).
This research looks at the intersection of the concepts of re-useable
enterprise models (i.e. reference models) and the provision of IT services. In
particular, this research is focused on the business processes of an IT
service provider (an organisation providing IT services), mirroring the trend
toward process management (Hammer 1990; Davenport 1993; Hammer and
Champy 1993).
This research is part of the parent project entitled “Process-oriented
Administrations of Enterprise Systems” (Rosemann 2002) which is based on
the concept that the same principles of business process management, that
have aided process management in industries such as manufacturing and
retail among others, can be applied to the provision of enterprise system
(synonyms are enterprise resource planning, information systems) services.
The research team consists of Associate Professor Dr Michael Rosemann as
chief investigator, Associate Professor Dr Glenn Stewart, and Mr Craig
Huxley and the author, Chris Taylor. Huxley’s work centred around using the
Masters Thesis Chris Taylor
QUT Page 3
Balanced Scorecard (Kaplan and Norton 1992) to develop a method for
identifying and prioritising the processes on which an organisation should
conduct process improvement activities. His work is outlined in his Masters
thesis (Huxley 2003).
This research project is funded by the Australian Research Council (ARC) in
conjunction with an industry partner (REALTECH).
1.2 REALTECH Australia
REALTECH Australia is a subsidiary of REALTECH AG which has its
corporate headquarters in Walldorf, Germany. The company today is a global
entity with offices and subsidiaries in 11 countries. The Australian subsidiary
was established in 1997 and its headquarters are in Sydney NSW.
REALTECH (www.realtech.com) specialises in development of application
and system management software, with the current product suite named
theGuard! The product suite covers the whole range of IT service provision,
including IT service level management, service support, application and
network management and inventory management.
REALTECH also offer consultancy services in all the areas of IT service
provision with specialisation in complex ERP systems.
Their interest in the research project was to leverage current knowledge and
improve its internal practices as well as the practices of its clients with
respect to IT service provision.
1.3 Research Objective and Research Questions
The research objective for this research is to apply the concepts of
reference models to the IT services domain in order to design a partial
business process reference model for IT service provision.
In reaching this research objective several issues are raised, and form the
research questions of this thesis.
Masters Thesis Chris Taylor
QUT Page 4
The overarching question is:
RQ1. What is an appropriate business process reference model for IT
service provision?
In order to answer this question several methodological questions need to be
answered. They are:
RQ2. What types of reference models exist?
RQ3. How can business process reference models be used for
process management?
RQ4. What characterises the quality of a business process reference
models?
And,
RQ5. How can reference models be produced?
The following pages of this chapter outline the structure of the thesis, which
is aimed at providing answers to these questions.
1.4 Thesis Structure
This thesis is essentially on two topics. Firstly, the driving topic of the thesis
is to explore, and develop where necessary, the ideas and concepts of
reference models in a generic sense, i.e. the answers to the research
questions 2-5. The second, topic is the application of these ideas and
concepts into a specific domain, that of IT service provision in order to
develop a business process reference model for IT Service Provision. The
thesis is structure generally follows this sequence, first the methodological
aspects of reference models then the application of these aspects to the IT
service provider domain.
The thesis commences with the exploration of reference models in a general
sense. This section examines, from a theoretical level what they are, what
defines their quality, how they can be created, and how they can be used.
Masters Thesis Chris Taylor
QUT Page 5
Within this section is the Literature Review, which explores the existing
literature on reference models in general.
The second topic of the thesis, IT service provision, the domain of the
reference model to be designed, is also explored next. The literature on
existing models and taxonomies of IT services is presented in the second
part of the literature review.
Chapter 3 introduces the research methodologies and design which are
employed in this research.
The next two chapters explore concepts and ideas of reference models,
developing theories about the quality and development of reference models,
theories that were not found in the existing literature. Following the
exploration of quality in reference models in chapter 4, chapter 5 provides a
theoretical basis for the development of reference models and a procedural
model for designing reference models is proposed and justified.
The topic of the thesis next becomes focused on the domain of IT service
provision, as the procedural model for reference model design is applied to
develop a partial reference model for IT service provision, which is outlined in
Chapter 6. During this design procedure, the model is tested in several
organisations, which is described in Chapter 7, to provide insight into the re-
use of the reference model and also to identify and integrate improvements.
Following the testing of the models, a discussion follows in Chapter 8. This
chapter discusses both topics, on one hand the IT service provision
reference model itself, and on the other hand provides an evaluation of the
procedure used to derive the model.
Chapter 9 summaries the thesis and main findings and provides a direction
for future work, focusing on the methodological aspects of designing and
applying reference models.
Masters Thesis Chris Taylor
QUT Page 6
Chapter 2: Literature Review
2.1 Introduction
As outlined in Chapter 1, the focus of this research is on reference models
and in particular business process reference models for IT service provision.
This chapter summarises the existing literature regarding the design, use and
purposes of reference models. It also highlights the different types of
reference models as mentioned in the literature, and then looks at the
literature surrounding the domain of this research: IT service provision.
2.2 Reference Models
This section looks at the definition of a Reference Model, the theoretical
underpinnings of reference models and presents a reference model lifecycle
that is used to position the various discussions of the thesis.
2.2.1 Definition
There is no one widely accepted definition of a reference model. The
confusion over the exact meaning of the term means it can be, and has been,
applied to many collections of information. The goal of this section is to arrive
at an integrated definition using various definitions, both explicit and implicit,
found accompanying or describing existing self-titled ‘reference models’.
This discussion begins by examining the words that make up the term in the
dictionary sense.
There are several definitions in the Oxford English Dictionary Online for the
term “reference” the most relevant to its use in the context are presented
here:
“The act or state of referring through which one term or concept is
related or connected to another or to objects in the world; also as
objective reference, and attrib. as reference class, property.
The process by which or the extent to which an individual establishes
a relation with elements in society as a standard for comparing status
and values
Masters Thesis Chris Taylor
QUT Page 7
book, etc. of reference, one intended to be, or suitable for being,
referred to or consulted. for reference, for the purpose of consulting or
being consulted.
a group to which a person may or may not belong but which he,
perhaps subconsciously, refers to as a standard in forming his
attitudes and behaviour
In extended scientific and technical use denoting an object, property,
value, or the like, used as a basis for comparative measurement or
standardization.” Oxford English Dictionary Online
Of particular interest in this thesis are the concepts of comparison,
consultation, and standards.
Similarly the most relevant dictionary definitions of the term “model” are:
“An archetypal image or pattern.
A simplified or idealized description or conception of a particular
system, situation, or process, often in mathematical terms, that is put
forward as a basis for theoretical or empirical understanding, or for
calculations, predictions, etc.; a conceptual or mental representation of
something.
Style of structure or form; design, structural type; build, make.
Of systems, institutions, and other immaterial things.” Oxford English
Dictionary Online
The important concepts here are understanding and prediction, simplicity and
idealisation, and a description. Combining these two definitions in a
dictionary sense could provide the statement that a reference model is a
standard description that allows for comparison. Of course, the definition of a
term using two words is often not as simple as combining the dictionary
definition of the two separate words. Therefore, other definitions are
explored.
Masters Thesis Chris Taylor
QUT Page 8
Several of the more widely known reference models are related to technical
computer system design. Presented here is a selection of definitions of a
reference model with respect to computer system design:
“A reference model is a generic manner to organise and integrate
system components.” (Zwegers 1998)
The ISO-OSI Reference model, one of the most widely known reference
models is an attempt to:
“provide a common basis … while allowing existing standards to be
placed into perspective … [and providing] a conceptual and functional
framework” (Anonymous 1998).
The reference model for XML design:
“provides a set of best practices that define the creation of XML
representations of standard business messages.” (Anonymous 2002
pVII).
In addition to these technically focused reference models, there is an
increasing list of reference models aimed at business processes in relation to
IT systems and also models solely focused on the business aspects, with
little or no relation to the IT supporting them.
Frank briefly defines a ‘generic’ reference model as: a representation of “a
class of domains” (Frank 1999 p.696) and that a reference model is:
“motivated by the search for general structures that can be applied to
numerous instances”. (Frank 1999 p.696)
Biemans continues with this description aspect:
“A reference model represents a system as an organisation in terms of
a structure of relatively independent, interacting, and in terms of the
globally defined tasks of these components.” (Biemans 1990 p.35)
Bernus adds:
Masters Thesis Chris Taylor
QUT Page 9
“[Reference Models] capture characteristics common to many
enterprises within or across one or more industrial sectors.” (Bernus et
al. 1998 p6).
SAP’s R/3 reference model, in part is described as a “graphical depiction of
industry reality”. . The US DoD states that it’s reference model can be used
“when a consistent and extensive … terminology is required” (Anonymous
2001 p7).
Misic and Zhao describe the reference model as a problem solving tool
stating that they are descriptions of:
“the standard decomposition of a known problem domain into a
collection of interrelated parts, or components, that co-operatively
solve the problem” (Misic and Zhao 2000 p484).
This use of a reference model for improvement is also reflected in Mertins
and Bernus, when they say a reference model is a:
“reusable enterprise model of common business processes designed
to improve efficiency and planning of new, or redesigning of existing,
processes.” (Mertins and Bernus 1998 p616).
Helbig pushes the concept of a reference model beyond the mere depiction
of reality into the shaper of reality saying:
“reference models are general models that can be used as a standard
for a whole branch, for a group of enterprises or for certain areas…
they are normative and implicit suggestions for a class of enterprises
they are aimed at.” (Helbig 1999).
This provides a view on reference models that they are suggestive in their
content, providing a picture of what and how something should be.
Rosemann also states that reference models are inherently suggestive, as
they “formalise recommended practices for a certain domain” [emphasis
added] (Rosemann 2002).
Masters Thesis Chris Taylor
QUT Page 10
Combining the aspects of problem solving and standardisation, the process
reference model definition provided in the SCOR Overview Booklet, reads in
part:
“a standard description … a framework of relationships, management
practices that produce best-in-class performance standard.”
(Anonymous 2002 p5).
This definition moves from the merely suggestive nature of a reference
model, to the extreme that a reference model contains the best standard.
Based on the reoccurring themes in Reference Model descriptions and
definitions, the following serves as the definition for this research:
A Reference Model is an abstracted depiction of reality (either
current or foreseeable, that serves as a standardised or
suggestive conceptual basis for the design of enterprise specific
models, usually within a like domain.
The terms “standardised or suggestive” were chosen because they represent
the most balanced view of the mentioned definitions. The words lie between
the extremes of the purely descriptive nature, inferred by words like
“common” (Bernus et al. 1998 p6) or “depiction of reality” , and the assertive
terms such as “best-in-class” (Anonymous 2002) or “recommended”
(Rosemann 2002).
An “enterprise specific model” is a model of a part of, or particular view of, an
enterprise or organisation and the comment “usually within a like domain”
reflects the statements of “for a class of” (Helbig 1999), “a known problem
domain” (Misic and Zhao 2000 p484) and “within or across one or more
industrial sectors” (Bernus et al. 1998 p6).
There are many other terms surrounding “Reference Model” and the next
section examines some of them.
Masters Thesis Chris Taylor
QUT Page 11
Overlapping Terms
Reference Models are also referred to in the literature as Partial enterprise
models, or Type I Reference Architectures. These terms have the same
meaning (Bernus 1998 p7). In the same paper, Bernus also identifies the
terms reusable, paradigmatic and typical models as synonyms.
Other terms that are often used as synonyms for reference model are
Technical Report Type II (Anonymous 2002), Stereotypical models (Chin et
al. 1989), Blueprint (Curran and Ladd 2000) and other terms such as
Framework or Guidelines.
Fettke and Loos also identify universal, generic and model patterns as
synonyms (Fettke and Loos 2003).
The term Architecture is also used with respect to reference models. With
respect to conceptual modelling architecture is defined in computing terms
as:
“The conceptual structure and overall logical organization of a
computer or computer-based system from the point of view of its use
or design; a particular realization of this.” Oxford English Dictionary
Online.
Zachman uses the term architecture as a classification for types of models,
for example the system model-function box in the Zachman framework gives
the example “Application Architecture” (Zachman 1987).
The guide to a reference model designed by the US Department of Defense,
states:
“the [reference] model is not … an architecture … [it] is an aid to
developing architectures”. (Anonymous 2001 p3).
Using the most liberal interpretation of ‘model’ i.e. a simplified description of a
situation (c.f. the definition provided in 2.2.1 of a model), an architecture can
be seen simply as a type of model. Indeed the definition of architecture as
above indicates that the architecture can be an instantiation, (i.e. a
Masters Thesis Chris Taylor
QUT Page 12
realisation), of a conceptual structure. The understanding that a reference
architecture is an instance of a reference model is reflected in (Zwegers
1998):
“A reference model specifies the general structure of the system …A
designer might use a certain reference model to specify an
architecture.”
However, as seen in this extract from (Anonymous), the terms model and
architecture have been used interchangeably:
“A most significant example of a reference architecture in information
systems is the Reference Model of Open Systems Interconnection
(often called the seven layer model) developed by ISO in the 1970’s.
This model has underpinned ...” [emphasis added] p1.
Others have used the term architecture to indicate a framework for possible
model types. This perception is particularly popular in the conceptual
modelling area, as distinct from technology development specifically.
Examples are the Zachman Framework, ARIS, CIMOSA, GRAI/GIM, IEM,
PERA (Bernus 1998 p7).
This thesis does not attempt to create a definition of ‘reference architectures’
and this discussion is only to acknowledge that some uses of the terms have
significant overlap in meaning. The definition provided in the previous section
will constitute a reference model for the purposes of this thesis.
2.2.2 Theoretical Justification
The justification behind a reference model relies partially on the assumption
that total variance of information and the means of expressing it at an
instance level is not completely rational. In other words, it seems simply that
representing every single situation with a different depiction is excessive and
that re-use of some of the depictions is justifiable (Frank 1999). The view
taken is that, either by re-using existing common representations, or by
developing new representations that can be applied in many instances, a
common representation can be formed, at least partially, about any domain.
Masters Thesis Chris Taylor
QUT Page 13
In the words of Frank, “This method of reducing the variance by introducing
new common concepts to handle information would not necessarily cause
dysfunctional effects.” (1999 p696) Frank goes further to say that if this
method of reducing variance, i.e. re-using depictions, were applied correctly
that it could contribute to efficiency.
Indeed the basic concept of re-use and abstraction and generalisation of
information used in reference models is already used in almost every model
or representation. Using a constructivist approach, due to the fact that a
model is simply a complexity reduced snapshot of real world concepts, it
stands to reason that different concepts, stripped of their non-relevant
information, can be generalised and represented singularly (i.e. with one
single symbol or word or concept). Every time this singular representation of
different real-world concepts is applied there exists instances of re-use and
the manifestation of generalisation. It is this same concept of generalisability
and re-use, albeit at a higher-level, that provides the justification for the
creation of reference models.
As defined above a reference model is one tool or method that can be
applied for the re-use of depictions. The next chapter proposes a simple
reference model lifecycle to situate the reader for further discussion.
2.2.3 Reference Model Lifecycle
Although no lifecycle for business process reference models has been found
in the literature the use of a simple lifecycle aids in the discussion of
reference modelling and is presented in Figure 1. It shows the lifecycle of the
reference model, which is simply its Design (to be examined in Chapter 5)
and Use (in Chapter 2.6), and also shows that the reference model has
classification characteristics (in Chapter 2.3), and Quality attributes (which is
examined in Chapter 4).
The proposed lifecycle is simply a tool for discussion in this thesis, but could
be extended to include aspects such as maintenance of the model. Due to
the way the thesis has been structured, which attempts to present existing
theories, then develop and apply them to the ITSP domain, the thesis
Masters Thesis Chris Taylor
QUT Page 14
structure will not follow the reference model lifecycle. Also the research on
the quality of reference models, is designed to provide input into how a
reference model should be designed.
Figure 1: Reference Model Lifecycle and Characteristics
The next section examines the classification characteristics of reference
models.
2.3 Classification of Reference Models
As made apparent by the literature review conducted for this study, varied
collections of information have been described as reference models. It is
helpful when studying a field to provide a system of classification. The
classification of reference models could be useful for a variety of purposes.
For example, an end-user of the reference model could use a list of classified
reference models in order to select the most appropriate for the intended use
of the model. The classification could also aid in the design of new reference
models, by making the relevant characteristics of the model explicit at the
time of design. Similarly a classification scheme could allow for the
identification of ‘holes’ in knowledge and also allow for comparison and
improvement or integration or consolidation of other like models (Fettke and
Loos 2003). The classification scheme could also provide a basis for
comparison of relative quality by allowing models to be compared with others
of the same type. Little work has been done to compare reference models,
Design Reference
Model
Reference
Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 15
and in particular provide assistance in the selection of reference models
(Misic and Zhao 2000).
This section focuses identifying the classification characteristics of a
reference model as shown in Figure 2.
Figure 2: Chapter 2.3 in relation to the Reference Model Lifecycle
2.3.1 Characteristics from Literature
This section is focused on providing that classification system, and examines
the different types of a reference model as seen in the academic and industry
literature. There has not been much work on the classification of reference
models, in fact only one work has been found, which is described in detail.
Other work on the classification of information models (or conceptual models)
is described after this. Rather than describe each mentioned characteristic in
detail they are only mentioned, then summarised and grouped into a table,
from which the classification scheme used for this thesis is derived and
explained in detail.
Fettke and Loos
The classification scheme itself is based on Fettke and Loos’ 2003 scheme
(Fettke and Loos 2003). A brief high-level discussion of their classification
scheme is presented here.
Design Reference
Model
Reference
Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 16
Fettke and Loos grouped their characteristics according to two dimensions,
specificity and domain dependence. The first dimension describes whether
the classification characteristic is dependent on the context (or domain) of the
model. Domain independent characteristics describe only the formal aspects
of the model, e.g. language, while dependent characteristics define the
domain of the model. Therefore the domain dependence can be either
Dependent or Independent.
Fettke and Loos defined the following classifications; Domain Dependent,
which included Functional Area (e.g. Research and Development, Sales,
Inventory, Production) and also Economic Activity (e.g. Manufacturing,
Trade, Health). The other classification group the Domain Independent
dimension consisted of View (Structure, Behaviour, Function) and Language
(ERM, Function Tree, EPC, Object-Oriented Approach). The second
dimension, describes the Specificity, which is whether the characteristic is of
a group of reference models, or whether the characteristic is of a single
reference model. The specificity of the characteristic can therefore be
“General” or “Specific”. In Fettke and Loos’s paper, this dimension is largely
ignored in their final classification scheme.
Subsequent correspondence with the author (Fettke 2003) indicates that the
reason for this is that examination of Independent-Specific characteristics is
highly dependent on the language selected and that examination of
Dependent-Specific characterisations is highly dependent on the chosen
domain making it difficult to generalise for inclusion in a classification
scheme.
The two aspects of Fettke and Loos’s dimension of Domain Dependence
relate neatly with Lindland et al.’s view of the quality of models, which is splits
the entities surrounding modelling into the model itself, the model domain,
the model language and the model audience (Lindland et al. 1994). Fettke
and Loos’s domain dependent characteristics are related to Lindland et al.’s
“domain”, while domain independent are related to the “language”. There is
no corresponding dimension to the Lindland et al.’s audience in Fettke and
Masters Thesis Chris Taylor
QUT Page 17
Loo’s framework, which is reasonable because the characteristics of a model
are independent of any potential audience.
The classification is summarised in Table 1.
Table 1: Fettke and Loos (1999) Reference Model Classification Scheme
Rosemann 1995
Rosemann’s “Morphologic box of information modelling” in (1995) find
reference) outlines 5 characteristics. They are Description view,
Decomposition Layer, Requirement for validity, Individuality of content and
Level of abstraction.
GERAM
The General Enterprise Modelling concepts defined by Bernus are similar to
the “View” category in the other 2 classification frameworks (Bernus 1998):
“Human oriented concepts cover human aspects such as capabilities,
skills, know-how and competencies as well as roles of humans in the
enterprise organisation and operation…
Process oriented concepts: They deal with enterprise operations
(functionality and behaviour)…
Technology oriented concepts, deal with various infrastructures used
to support processes and include for instance resource models
Masters Thesis Chris Taylor
QUT Page 18
(information technology, manufacturing technology, office automation
and others)” (Bernus 1998 p7).
Rosemann 2002
Rosemann (2002) also discussed the classification of reference models
describing several classifications. Including View, Granularity, Scope,
Internal/External, Availability of Model, Textual explanation, Use of
Guidelines, Benchmark data, Explicit alternative scenarios, User Group,
Purpose.
The authors above often do not give a detailed definition of their classification
characteristics. Using the information and terminology the authors above
have presented, the concepts have been grouped as shown in Table 2.
Masters Thesis Chris Taylor
QUT Page 19
Origin Proposed Characteristic
Type Fettke '03 Rosemann '95 GERAM Rosemann ‘00
Description view
Entity Model Content Views
View View
View
Generic Enterprise Modelling Concepts
Language Formality
(Subsumes) Language
Granularity Level of Detail
Decomposition Layers
State Requirement for validity
Functional Area Functional Area Scope
Internal/External Economic
activity Economic activity
Tool Support Availability of Model
Extended Content
Textual explanation,
Use Guidelines, Benchmark data
Readiness for Use
Explicit alternative scenarios
(Partially) User Group
Purpose
Level of Abstraction
Genericity Dimension
Table 2: Existing and Proposed Model Characteristic Type Overlaps
In Table 2 the proposed set of classification characteristics, shown in the left
column, is compared to the existing literature. The only one not covered by
the proposed new characteristics is Rosemann’s “Level of Abstraction”
Masters Thesis Chris Taylor
QUT Page 20
similar to the dimension of “Genericity” (not shown in table) in the CIM-OSA
framework (Anonymous 1993; René Gaches Year Unknown) or GERAM’s
Genericity dimension and Reyneri’s “Genericity” in (Reyneri 1999). These
concepts are similar to Malone et al.’s dimension of “Type” which indicates
the degree of generality or specificity of the model (Malone et al. 1999).
From the definition of a reference model proposed above in 2.2.1, i.e. a
depiction of reality, then the reference model cannot be a meta-level model
and it cannot be a model with only one intended use (usually a characteristic
of a model of one specific situation as opposed to a generalised model). This
confines a reference model to a certain range of genericity. This could
include a reference model used specifically for one organisation to a
reference model for a generic enterprise. This range of genericity is covered
in the functional and economic activity categories as explained below.
Another classification characteristic that has been proposed for reference
models, and is not shown in the table nor been addressed specifically is
“perspective”. Perspective has been defined as the combination of the target
user group and the purpose of the model (Rosemann and Green 2000).
Purpose can have a many to many (n:m) relationship with the use of the
model. That is a reference model may have many purposes and it may be
put to many uses, with no necessary overlap between the two sets. A
reference model might be designed for developing workflow within an
organisation, but it could be used for training or building an application to
support the workflow. Only the reference model’s tangible and intrinsic
characteristics may be relevant when selecting a reference model. Purpose
is deliberately left out of the classification scheme because it is not an
intrinsic characteristic of the model. The effect of the choice of purpose made
during the design of the reference model, is most evident in the Focus
(Technical, Business, Application) and View (Data, Function, Process,
Resource) characteristics of the resulting model. These characteristics are
discussed in detail below.
Masters Thesis Chris Taylor
QUT Page 21
The target group of the model is similar to the purpose of the model in that
the target group may differ from the actual audience or audiences of the
model. The target group influences the characteristics of the model being
designed, particularly the Focus and the Language formality (Natural, Meta-
Model or Ontological Theory) as discussed below.
2.3.2 Derived Classification Characteristics
The next section outlines the characteristic types then the individual
characteristics proposed within these types, with a short description.
View
The characteristic “View” is already widely discussed in literature (e.g.
Bernus 1998; Rosemann 1998; Fettke and Loos 2003). There are 4 common
views often mentioned in literature as early as 1992 (Curtis et al. 1992)
although they were originally termed “perspectives” although this term has
evolved to now mean the union of the intended audience and the purpose of
a model (Rosemann and Green 2000). The most common views include,
functional, behavioural, organisational and informational.
Functional
The functional perspective describes what activities are performed. Terms
such as tasks or activities are often used when depicting functions. Functions
are often used in both behavioural and resource views, however for the
purposes of this thesis, the functional view is used to denote those models
that only examine the functional aspects of a situation without regard for the
sequencing or organisational or resource implications.
Behavioural
(or Process or Control): The behavioural view captures the rules about
when/why/how functions should occur. It includes rules and often outline
when the function is performed (sequencing, conditions, loops). It includes
concepts such as sequencing, decision paths, conditional branching loops
etc. In other words, the behavioural view defines how functions are employed
in response to a stimulus (Anonymous Year Unknown). Some authors e.g.
(Scheer 1998) put process under this heading, although strictly the process is
Masters Thesis Chris Taylor
QUT Page 22
a combination of the functional and behavioural views. This strict
interpretation is not adhered to in this thesis and the terms behavioural and
process is used interchangeably for the sake of simplicity and adherence to
common usage.
Organisational
(or Resource) This view shows the resources that are involved in the model.
It is often used for human resource planning but could also involve computer
hardware of software components or other consumables.
Informational
(or Data) This view looks at the temporal concept of data or information. It
can be a depiction of the existence of data, how the individual datum relates
to each other and how it is structured. These models are applied to areas
such as database management, content management or knowledge
management.
Dependent on the goals of the model, other views can be added (e.g.
motivation, technical), these 4 however, are the most commonly mentioned in
literature.
Language Formality
From Fettke and Loos, the language describes the modelling technique,
notation or grammar of the model (Fettke and Loos 2003). Numerous
modelling languages exist for any particular view, and, the language of the
reference model is an important consideration particularly, with respect to
using a reference model in an organisation which already has a standard
modelling technique or experience in a particular language.
Reference models, like enterprise specific1 models come in many languages.
To avoid an endless list on languages, the classification scheme presents
only three categories, Natural, Meta-Model and Ontological Theory
1 The term enterprise specific model is used to indicate a model of a specific situation, it is acknowledged that not all of these models is of an enterprise, but most of the models discussed in this thesis is related to an enterprise.
Masters Thesis Chris Taylor
QUT Page 23
describing the formality of the language. Models that are a mixture of the
following are categorised by the most prevalent language type in the model.
Many reference models may not explicitly state that they are built with a
meta-model or ontological theory in mind. In this case the classification must
be made on the appearance of the use of a meta model or theory. Generally
commonly used, “standardised” languages (such as Petri Nets, Event Driven
Process Chains, UML) at least have a meta model.
From GERAM the three language formality characteristics are:
Natural
This characteristic describes models that are written in natural languages.
The most prevalent of these is plain English.
Meta-Model
Models that present, or are obviously based on a formal description of the
modelling technique are called meta-model type models. A meta-model
consists of the grammar, rules and symbols that define the modelling
language. It can be thought of as the description of the technique and syntax
to which a model should conform. This should include models that are only
implicitly based on a meta-model.
Ontological Theory
These models present both a meta-model and define the semantics of the
meta-model. That is, not only does the meta-model define the types of
symbols should be used and how these symbols can be connected, but it
provides a semantic meaning to the symbols and connections. This
effectively creates an explicit map between the real life constructs and the
symbols used to represent them. Typically these languages are presented
built in dedicated modelling tools.
Level of Detail
Similar to Rosemann’s “Decomposition Levels” and Zachman’s “Rows”, the
level of abstraction describes the level of detail that the model shows. It is the
same dimension as Malones’ “Uses/Sub-Activities” (Malone et al. 1999). The
levels are complete, intermediate and task.
Masters Thesis Chris Taylor
QUT Page 24
Complete
The complete level is designed for the highest level conceptualisation. It is
often used in the initial phases of a modelling or systems design project. It
provides the big picture of the concept under review. The complete level
deals with issues such as planning, opportunities and objectives. Broadly
speaking the complete level defines the purpose of the object. Using a
specific type of modelling (business process modelling) as an example, at the
complete level executives outline strategies and goals.
Intermediate
The intermediate level is a step down from the higher complete level
depiction. It provides more detail allowing segmentation of the entity. It shows
how the major components of the system fit together and inter-relate. Again
with the business process model example, at the intermediate level,
management in the business units sets direction for their organizations.
Task
The task level is the level at which further decomposition of the model is
inappropriate and comprises ‘discreet’ or ‘atomic’ units. An example of this
level is the level in application reference models that depict individual
transactions or screens.
Focus
This characteristic is a description of which parts of the domain are of interest
in the model. This characteristic type is an indication of the ‘real-world’
concepts that the model is trying to capture. It can be thought of as an
answer to the question “what is the model of?”. The characteristics provided
are Business, Technical and Application.
Business
A business model’s main goal is to describe a business concept, for example
how accounts should be structured or how business processes are executed.
While business models may be used to build applications or IT infrastructures
the majority of the content is based on issues to do with aspects involving
people and organisations as entities. These would be used by managers or
Masters Thesis Chris Taylor
QUT Page 25
operational staff among many others and often contain some form of
benchmark. An example of this type of model could be the insurance
reference model by KPMG (Scheer 1998).
Technical
A technical reference model’s main goal is to describe industrial, mechanical
or IT related concepts. This could be tangible concepts, such as routers and
computers or machines, or it could be intangible concepts such as
networking protocols. These would be used by technicians, IT staff and those
responsible for design or maintenance of the technical system being
examined. An example of this type is the OSI/ISO Model (Anonymous 2003).
Application
Application reference models are used to describe the functionality of the
specific application being modelled. An application model can be used as a
communication medium between IT staff and end users among other uses.
The application reference model’s usually serve as a tool to facilitate the
implementation of the system, typically an Enterprise System. The most well
known of these is the SAP R/3 reference model, it describes some of the
processes and functions that the R/3 system can support. Other examples
include the Baan or Siebel reference models.
State
Similar to Rosemann’s “Requirement for Validity” this characteristic type
describes exactly what state the model is representing. The two
characteristics given are Common Practice and Best Practice. Over time a
reference model may move between these categories (especially from Best
Practice to Common Practice), just as today’s innovative idea becomes
accepted and widely used. This can be loosely thought of as an answer to
the question “what time is this model depicting?” or “how state of the art is
the concepts in this model?”, with the common practice depicting now and
the best practice depicting sometime in the future. This is slightly different to
Rosemann’s Requirement for Validity because reference models are
developed at a generalised level, and hence company specific concepts,
such as the differentiation between a TO-BE model (i.e. usually the short
Masters Thesis Chris Taylor
QUT Page 26
term goal at the end of a particular project) vs. an Ideal model the ultimate
long term goal, are not appropriate.
Best Practice
The term ‘best practice’ is used to indicate the notion of the ideal end-state
situation, as typically perceived by experts in a particular field. It could be an
existing running system, or a combination of existing solutions and new
ideas. It is used here to indicate the perceived best solution for a particular
problem domain. The term “best-practice” is a highly subjective term, and is
often difficult and usually impossible to prove or disprove the validity of a best
practice claim. Other terms could be “better”, “good” or “accepted” practice.
Best practice incorporates a kind of ideal end state with no constraints on
resources, situations, implementation costs or politics. The best practice
might also be a representation of the minimum standard an organisation
should achieve.
Examples of best practice reference models e.g. ITIL/SCOR. Many reference
models claim to be best practice models (perhaps usually for marketing
reasons), particularly those which are not technical models.
Common Practice
Some reference models are an attempt to map the current practice (i.e.
common practice), for example SAP R/3 reference model, or the reference
model developed for university finances in Germany (Anonymous 2001).
Functional Area
The functional area describes the scope of activity modelled. The term
“functional” in this case refers to Taylor’s functional approach (Taylor 1911).
It describes the vertical types of activities that an organisation undertakes. A
reference model can represent one or many of these functional groups. It is a
domain dependent characteristic type, in that this classification is dependent
on the content of the model and not the representation of the model itself. It
answers the question of “where is this model set?” or “for what scope of
domain is the model valid?”. Again to avoid an almost endless list of
Masters Thesis Chris Taylor
QUT Page 27
possibilities, three characteristics are presented, Function Specific,
Enterprise and Inter-organisational.
Function Specific
A function specific model is when the model scope is limited to a specific
functional area (or multiple defined functional areas). Examples could be
human resources management, financial management, production etc. An
example is (Malone et al. 1999) which focuses on manufacturing.
Enterprise
An enterprise reference model is a depiction of a single organisation for
example the Process Classification Framework (Anonymous). It would
include all the functional areas for an economic activity or other grouping of
organisations under question. An example would be the SAP R/3 reference
model. The Enterprise reference model would be a superset of the functional
models.
Inter-organisational
An inter-organisational reference model captures the group of models that
represent inter-organisational concepts such as the supply chain, customer
relationship, virtual enterprises or communications. These have been
specifically designed to capture the interactions between organisations or
entities. Examples would include the Siebel model for Customer relationship
management or the Supply Chain Operations Reference Model (SCOR). The
Inter-organisational model could be a superset of the enterprise models.
Economic Activity
As proposed by Fettke and Loos (Fettke and Loos 2003) economic activity is
sometimes what could be called “Industry” but has a more defined meaning
with respect to the United Nations’ “International Standard Industrial
Classification of All Economic Activities” (Anonymous 2002). Again this helps
to answer “where is the model set” or “where is the model most applicable?”.
However, instead of focusing on functional classification it focuses on
economic activity. Three characteristics are presented; Organisation Specific,
Economic Activity Specific and General.
Masters Thesis Chris Taylor
QUT Page 28
Organisation Specific
Some reference models are designed for a particular organisation and are
then used to standardise or benchmark internal processes across instances
within the organisation, such as defining the financial reporting across
different regional offices of an international organisation.
Economic Activity Specific
Economic activity specific models are those that are defined for a particular
economic activity, indicating that their use may not be appropriate for use
outside that domain examples include Financial Intermediation, Public
Administration, Manufacturing, Hotels and Restaurants etc. examples would
be the various versions of SAP R/3’s “Industry Solution Maps”.
General
General reference models in this sense are not related to a particular
economic activity, an example being the Process Classification Framework
(Anonymous).
Tool Support
This indicates whether the model is available in a dedicated modelling tool.
The modelling tool is almost always a computer application and tool support
indicates whether the data file for use in this tool is available. The model can
be provided by the model producer, or a third party who takes the original
model and releases a version of the model in their toolset and language. This
can require some interpretation on the part of the third party. Of course some
reference models are not directly supported by any toolset. Similar to the
distribution characteristic type the same content may be offered with any of
the following tool support characteristics.
Producer supplied
These models have been provided in a data file by the producers of the
reference model. For example, the SCOR model supplied by the Supply
Chain Council.
Masters Thesis Chris Taylor
QUT Page 29
Third party supplied
Some models have been replicated (often with a degree of interpretation) into
the language of a modelling tool by a third party, often consultants or tool
vendors. For example, eTOM supplied by IDS-Scheer.
Extended Content
The characterisation describes whether the model comes with any model or
content ‘extras’. These extras are delivered in addition to the core of the
model. Three characteristics for this type are considered Implementation
Information, Run-Time Information and Model Explanation.
Implementation Information
A model exhibiting this characteristic would include guidance on the
implementation of the situation described in the model. This could come in
different forms such as an implementation methodology. Any information that
could be seen as helping the implementation of the situation in the model can
be classified as implementation information.
Run-time Information
This would include information or guidance in the form of case studies or
performance data for benchmarking. Run-time information is any information
that gives the user an extended view of the post-implementation situation or
challenges faced with running the proposed situation. Aspects of any training
aimed at linking the model to real life examples or explaining the content
would be included in this characteristic.
Model Explanation Information
Extended glossaries, meta-models, or explanations about how to read or
interpret the model would be included here. This extended model information
is aimed at facilitating the model understanding. Training in the language of
the theoretical concepts in the model would be included here.
Masters Thesis Chris Taylor
QUT Page 30
Readiness for Use
This characteristic gives an indication of how the reference model could be
derived2 (i.e. transformed into an enterprise specific model) or at least how
the reference model has been designed for derivation. Derivation is the
process of producing the enterprise specific model starting from the
reference model. It is unlikely that any reference model can be used without
some changes, and this characteristic is not an attempt to define whether
work will need to be done adapting the reference model before use; rather it
is an indication about what type of work is required to convert the reference
model into an enterprise specific model.
Single depiction
A model that is characterised as a single depiction depicts with a one to one
relationship to its domain. There are no built-time (or implementation)
alternatives presented and the model is presented in a “ready-to-run” off the
shelf manner after a certain amount of changes to fit the model to the
situation. This type of model has been referred to as a “prototype” model
(Bernus 1998).
Contains variants
A model that contains built-time variants has explicit alternative options for a
single situation, requiring a user to choose one or more of the explicit
alternatives before implementing the model.
Abstract
This type of model is similar to Bernus’s “Abstract” (Bernus 1998) reference
models or Loo’s “Generic Structures” (Loos et al. 1996). These are reference
models that are at a level of abstraction that they require the addition of
enterprise specific detail to produce the enterprise specific model. These
models require that the user “fill-in-the-blanks” (Bernus 1998) or require
2 The term “derivation” is used to describe the process where by a reference model is transformed into an enterprise specific model. It is similar to “adaption” from GERAM (Bernus 1998) or “instantisation” from the CIMOSA modelling framework (René Gaches Year Unknown).
Masters Thesis Chris Taylor
QUT Page 31
mapping to certain specific semantic meanings before they can be used. This
characteristic is related to the “Level of Abstraction” characteristic type. It is
subtly different however, because it is drawn not from the level of abstraction
of the model, but the difference between the level of abstraction in the
reference model and the expected abstraction level of the Enterprise Specific
Models. An “Abstract” model is specifically designed to provide a high level
framework for the development of the enterprise specific model.
2.3.3 Characteristics from a Review of Reference Models
Along with these previously defined characteristics a review of available
reference models reveals there are other important aspects of reference
models. As mentioned earlier this classification was designed in part to aid
selection of reference models in real life projects. With this in mind, a brief
examination of available reference models was conducted and several more
characteristic types have been added to the scheme. These were
Distribution, Ability to Edit, and Currency. For the purposes of classification in
this thesis these new characteristic types combined with the previously
mentioned are complete and necessary.
As above each of the characteristic types are explained then characteristics
are proposed and explained below.
Distribution
This characteristic type describes how the reference model is made
available. The characteristics are public domain, proprietary available for
sale, proprietary available through membership and proprietary not available.
Some reference models are available through multiple sources, for example
ITIL is a public domain reference model, but several third parties claim to
have ITIL modelled in their proprietary languages. These third party supplied
models are proprietary for sale. In this case a single source reference model
may be categorised in several distribution characteristics.
Public domain
A public domain reference model is one that is made freely available to the
public or pay only for the publishing costs.
Masters Thesis Chris Taylor
QUT Page 32
Proprietary not available
This characteristic describes those reference models that are not available
for the public in any form, often these are the intellectual property of
commercial organisations such as consultancy firms who would see that
information as their competitive advantage. Such models may become
available to the clients of such organisations. An example of this is the IT
Service Management reference model from HP.
Proprietary available for sale
These models are available for purchase from the developers or distributors.
An example of such as developer is IDS-Scheer based in Germany.
Proprietary available through membership
These models are available after obtaining membership (usually for a fee)
with an organisation, typically a body representing a particular industry or
area. An example of this method of distribution is the SCOR reference model
which is available to members of the Supply Chain Council.
Ability to Edit
Related to the tool support is whether the reference model is able to be
edited or modified in its original form.
Locked
Locked models are not modifiable in there original form. This would include
reference models that are supplied in hard-copy only, or media such as a
locked adobe acrobat file.
Open
Open models are modifiable in their original form, assuming access to the
appropriate tool is available.
Currency
Currency is an indication of the timeliness of the model content, and whether
the content of the model is updated.
Masters Thesis Chris Taylor
QUT Page 33
Living
Living models are reviewed and updated with a degree of regularity. This can
be done through inputs from users of the models, surveys or other review
mechanisms. Upgrading a reference models faces similar challenges to the
design of new models (Design is addressed in Chapter 5).
Static
These reference models are those that were never updated, they are
released once in a final version.
Discontinued
These models are those that were once living models but the review and
update processes has been stopped. This is particularly evident in several of
the OSI reference models or the TeleSPICE model which received many
updates soon after its release, and has had no updates for several years.
2.3.4 Classification Summary
Presented in Table 3 is the proposed classification scheme showing the
characteristic types drawn from the academic literature as described in Table
1, and the new characteristic types drawn from a review of existing reference
models.
Masters Thesis Chris Taylor
QUT Page 34
Characteristic Type Comb Characteristics
View M Data/ Information
Process/ Behaviour Function Organisational
Language Formality 1 Natural Language Meta-Model Ontological Theory
Level of Detail M Complete Intermediate Task
Focus 1 Business Technical Application
State 1 Common Practice Best Practice
Functional Area 1 Function Specific Enterprise Inter-organisational
Economic Activity (Industry) 1 Org. Specific EA Specific General
Tool Support N Producer supplied Third party supplied Public Domain
Extended Content N Implementation Run-Time Model Explanation
Readiness for use 1 Single depiction Contains variants Abstract
Distribution M Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit 1 Locked Open
Currency 1 Living Discontinued Static
Table 3: Classification of Reference Models
The column marked with “Comb” denotes the allowable combinations of the
characteristics.
“M”: Indicates that any combination may be possible, but at least one (i.e.
1,m)
“N”: Indicates any combination may be possible (including none) (i.e. 0,m)
“1”: The characteristics are mutually exclusive. (i.e. 1,1)
These combinations are a guide only, as it could be argued for example that
related models presented in languages of different formalities are separate
reference models, or simply parts of the single reference model. Also
characteristics of the model may change over time.
Masters Thesis Chris Taylor
QUT Page 35
This scheme simply suggests the relationship between the characteristics
within a particular characteristic type to aid in the application of the
classification framework.
This scheme has been drawn from literature and derived through the author’s
work and may not be exhaustive.
2.3.5 Example of Reference Model Classifications
Supply Chain Operations Reference Model
The Supply Chain Operations Reference Model (SCOR) has been developed
by the Supply Chain Council, a non-profit industry organisation focused on
supply chain issues. It contains descriptions of the processes of a typical
supply chain, based mainly around manufacturing industries. It uses a series
of arrows and boxes with specific naming methods and so falls into the Meta-
Model level of language formality, however there is no meta-model presented
(only implied). The level of sophistication of the modelling technique is very
low.
The nominal focus is on the business process of an organisation, however
the model is really only a functional model, with some dependencies on
information and resources (i.e. inputs and outputs) shown, but no business
rules for how or when the functions are executed. It attempts to highlight best
practice. Due to the nature of its domain the model is an inter-organisational
model, mapping the process through 5 organisations (from the suppliers’
supplier through to the customers’ customer) through a supply chain.
Although it claims to be industry independent the model only makes sense in
a manufacturing or distribution industry (i.e. not a service industry). More
specifically, it is most applicable to high turn over manufacturing or
engineering environments.
The main advantage of this model is the identification and definition of supply
chain performance metrics, giving the model extended “Run-time” content.
The model contains process elements that can be plugged into the model,
hence the model is of the “Contains Variants” Readiness for use type.
Masters Thesis Chris Taylor
QUT Page 36
The model is accessible through membership to the supply chain council,
and comes in two forms, either a document, or in a dedicated tool, the
different versions respectively locked and open.
The model at this stage is living and the description taken from Version 5.0 .
From the description above the following classification Table 4 is derived.
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 4: Example reference model classification of SCOR
This provides an example of how the characteristics are derived from an
examination of the model. The same straightforward method of assigning
characteristics was applied throughout the rest of the thesis, and will not be
explained again.
Masters Thesis Chris Taylor
QUT Page 37
2.4 Typical Applications of Reference Models
There have been many uses touted for reference models. To provide a
structure for discussion this section is broken into the ‘focus’ characteristic as
described in the previous section. The focus characteristic categorises what
the reference model is describing, and hence can a framework for analysing
how the reference models are used.
Technical Reference Models
One extensive use of reference models has been in the area of information
technology. Technical reference models generally provide a conceptual
framework for development of technical systems. Zwegers states that
reference models “serve as a point of departure for the design of a large
number of systems in a specific application area”. (Zwegers 1998).
Examples of technical reference models include many from the ISO,
including ISO/IEC 11072:1992 Information technology outlining computer
graphics standards or the ISO/IEC 14662:1997 Information technology -
Open-EDI reference model which describes the interfaces and standards of
Electronic Data Interchange. Others include the Workflow Reference Model
(Hollingsworth 1995) which defines the standard interfaces for a workflow
management system.
One of the most commonly referenced technical models is the ISO-OSI
reference model. It is provides a framework for the development of
communication systems and the seven layers depict the different elements
that make transform the communication from the human interaction to the
physical transmission and back up again to the human interface. It is
interesting to note that this is not the backbone of the most commonly used
electronic communication – the Internet (which uses TCP/IP). Despite it not
actually being widely used the reference model’s ability to be able to
effectively and simply define a complex situation has made it popular. This
highlights one of the uses of a reference model, to display complex ideas in
such a way that it is easily understood and conceptualised.
Masters Thesis Chris Taylor
QUT Page 38
What these reference models have in common is that they attempt to provide
an accepted conceptual or technical standard. This standard ensures that the
products developed under the standard can fit neatly into their respective
roles. They act as input for requirements engineering ensuring that vendors’
products conform to the relevant standards.
Application Reference Models The use of reference models is not only confined to the technical side of IT.
Recently several application reference models have been proposed that deal
with how IT systems can be used, particularly business process models of
enterprise systems (e.g. SAP R/3, Siebel or Baan Reference Models).
One of the major aims of application reference models is to aid with the
configuration and implementations of the applications they describe. They
help to align the application to the business requirements and can be used to
describe the supported business processes, data structures, architectures or
configuration alternatives of the applications. Typically the process reference
models of application models describe the human-application interface of the
application.
Application reference models also include models of business or human
interactions which are used to design or configure IT systems. Examples of
this type of reference model can also be found in the area of web services
and e-market places.
Business Reference Models
Business reference models are mainly focused on describing business
concepts, such as business processes, structures, financial arrangements
etc. An example is the Process Classification Framework (Anonymous) which
provides a generic functional decomposition for a commercial operation, or
the teleSPICE reference model (Emam et al. 1998; Anonymous 2002) which,
in part, describes the processes one should complete when developing
software.
Business Reference Models are useful to:
Masters Thesis Chris Taylor
QUT Page 39
• Communicate “best/common/accepted” practices
• Structure a performance measurement framework
• Encourage reuse in multiple instances and consistency in model
efforts
• Provide guidance or a template for modelling efforts
• Facilitate the classification, evolution and comparability of models by
creating a standard terminology
They provide assistance in
• Business Process Management (BPM)
• Knowledge Management
• Process Benchmarking and Performance Measurement
(Anonymous 1998; Bernus et al. 1998; Curran et al. 1998; Lowe and Webby
1999; Misic and Zhao 2000; Kaplic and Bernus 2001; Anonymous Year
Unknown)
As described in Chapter 1, of particular interesting in this research is
business process modelling and business process reference models. The
next sections look into both of these areas.
2.5 Business Process Modelling
A conceptual model is a simplified depiction of an extraction of reality (Curtis
et al. 1992). The information presented in the model is, in the model creators’
opinion, all the relevant information needed to fulfil the purpose of the model
(Lindland et al. 1994; Davis 2001). Ideally the model should contain only the
minimum required in order to fulfil its purpose and hence “excludes much of
the world’s infinite detail” (Curtis et al. 1992). This is done in two steps, firstly
minimising the scope of the model, focusing on only the relevant objects, and
secondly by depicting this defined scope in a minimalistic manner. By doing
this it would comply with Albert Einstein’s suggestion to “make everything as
Masters Thesis Chris Taylor
QUT Page 40
simple as possible, but not simpler” (exact source unknown). Modelling in
one form or another has taken place since the first caveman drew on the
wall. Several formal modelling techniques were developed with the
introduction of IT, as precise definitions are needed when working in the field.
Data modelling was one example that many organisations used.
In the 1990’s many organisations implemented large and complex enterprise
systems, including Enterprise Resource Planning packages (ERP). The
traditional focus on modelling data flows and transformations was not
sufficient to capture the increased use of IT beyond data processing into
areas of communications and co-ordination (Curtis et al. 1992). At the same
time, popularised by Hammer in 1990 (Hammer 1990), the change from
Taylor’s (Taylor 1911) scientific management principals to the process
perspective marked a dramatic shift in business management.
This deficiency in modelling techniques and the contemporary popular focus
on BPR increased the need for a process modelling. A process (used often
synonymously with business process) can be defined as:
“a self contained, temporal and logical order of those activities, that are
executed for the transformation of a business object with the goal of
accomplishing a given task” (Green and Rosemann 2000).
“the definition of the tasks and the sequence of those tasks necessary to
deliver a business function” (Davis 2001)
“a set of partially ordered steps intended to reach a goal” Humphrey and
Feiler (1992) in (Curtis et al. 1992), or
“a structured, measured set of activities designed to produce a specified
output” (Davenport 1993).
Hence process modelling (or business process modelling, BP Modelling) can
be defined as:
“the capture, documentation and analysis of [business processes]” (Davis
2001 p2), or
Masters Thesis Chris Taylor
QUT Page 41
“an abstract description of an actual or proposed process that represents
selected process elements that are considered important to the purpose of
the model” (Curtis et al. 1992 p.76).
The working definition for this thesis is:
Business process modelling is the activity of capturing and
analysing the important aspects of a business process.
In the early stages of process modelling the focus was on the use of models
to describe and implement IT based solutions. Business process modelling
differs significantly from the more traditional modelling techniques used in the
field of IT because “many of the phenomena modelled must be enacted by a
human rather that a machine” (Curtis et al. 1992 p77). Today, BP Modelling
is seen as part of Enterprise Modelling, and related upper-CASE (computer
aided software engineering) tools allow the integration of process design at a
conceptual level with the coding into Enterprise Systems, for the enactment
of these processes, and ultimately the monitoring of these processes (van
der Aalst and Kumar 2001).
Even though some models produced included both semi-manual and IT
executed steps the main emphasis was on the IT system, sometimes
excluding the solely manual steps. As BP Modelling matured the advantages
of using it in areas outside the IT domain became apparent. Today BP
Modelling has been used to map the entire process, as originally proposed
by Hammer, regardless of whom or what executes the function.
Various uses for BP Modelling have been summarised in Table 5 adapted
from (Rosemann et al. 2003).
Improvement of internal business processes
Improvement of collaborative business processes
Software development Design of Enterprise Architecture
Business process documentation Change Management
Workflow management Training
Masters Thesis Chris Taylor
QUT Page 42
Table 5: Examples of uses for Business Process Modelling
While parts of the remainder of this thesis may be applicable to reference
models in general, the term reference model is used to indicate business
process reference models in particular from now on.
2.6 Applying Business Process Reference Models
This section summaries and extends the literature with respect to the use of
business process reference models, as highlighted in Figure 3. In particular it
focuses on using business process reference models in Business Process
Re-engineering (BPR) projects, one of the major uses for business process
reference models. The section answers the third research question:
“How can business process reference models be used for process
management?”
Figure 3: Chapter 2.6 in relation to the Reference Model Lifecycle
Very few published works have mentioned how to apply a reference model.
Two exceptions, however, are brief mentions in Schlagheck and Scheer.
Schlagheck 2000 (in Fettke and Loos 2003) details four steps in the process
of applying a reference model:
Design Reference
Model
Reference Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 43
1. Problem Definition
2. Requirements definition
3. Reference model selection
4. Enterprise-specific model construction
Scheer (1999 p.94), while describing the steps used for individualising
reference models, gives 4 steps:
1. Determining the Vertical Market
2. Determining the Business Processes
3. Selecting the Reference Models
4. Adapting the Reference Models
These steps do not provide details on exactly how the reference model
should be used (other than the generic “adapt”) nor do they link it to any of
the activities in which modelling is embedded. Modelling is seldom an ends in
itself, but rather a means to some other ends. A common end is business
process management. Rosemann’s Business Process Lifecycle (Rosemann
2000), shown in Figure 4, outlines the steps in for business process
improvement (BPI) and business process management (BPM).
The remained of this section is an extension to the generic outlines provided
above on how to apply business process reference models.
Masters Thesis Chris Taylor
QUT Page 44
Figure 4: Business Process Lifecycle 3
Similar lifecycles or process steps can be found in McGowan and Bohner
(1993) and van der Aalst et al. (2003).
This lifecycle is used to discuss specific applications of process reference
models and also how they can be applied in each instance. To examine how
reference models can be used in the lifecycle, the applications discussed in
2.4 can be matched to the lifecycle steps.
Figure 5 shows how a reference model can be used in the business process
lifecycle. On the left is the process lifecycle, the middle column shows the
uses of the reference model and on the right are the steps that are needed to
select the appropriate reference model. Also shown in this figure are the sub-
sections of this chapter that describe the individual applications in more
detail.
3 Note: The original step of “process identification” has been broken into “process identification” and “process targeting” to aid in the discussion of the application of business process reference models.
Process Identification
Process Targeting
AS-IS Modelling
Analysis
TO-BE Modelling
Process Implementation
Process Execution
Monitoring/Control
Masters Thesis Chris Taylor
QUT Page 45
Figure 5: Business Process Lifecycle and related use of Reference Model
The following sub-sections describe the types of uses (Template, Scope
Definition and Targeting, Process Benchmark, Implementation Guide, Run
Time Information and Performance Benchmark) in more detail.
It should be noted that at present not many reference models have the
maturity to be used in all of the above mentioned ways. Indeed for a single
reference model to be able to be used in all the above ways, it would require
a significant investment in the model design.
Firstly the application as a template (high-level, AS-IS and TO-BE) are
described, secondly use as a process identification tool is examined followed
by the use as a process benchmark, as an implementation guide and finally
as a performance benchmark.
Select Reference Model
Process Identification High-Level Template (2.6.1)
Process Targeting Scope definition/ suggestions (2.6.2)
AS-IS Modelling AS-IS Template (2.6.1)
Analysis Process Benchmark (2.6.3)
TO-BE Modelling TO-BE Template (2.6.1)
Process Implementation
Implementation information (2.6.4)
Process Execution
Monitoring/Control Run Time
Information/Performance Benchmark (2.6.6)
Masters Thesis Chris Taylor
QUT Page 46
2.6.1 Templates
The most commonly discussed use for reference models is a template or
starting point for the design of enterprise specific models (Scheer 1994;
Mertins and Bernus 1998; Scheer 1999).
This subsection firstly examines the which parts of the reference model can
be re-used as a template, secondly it describes how the models can be
derived from the reference model, the derivation methods, and thirdly it
examines the individual applications of use as a template in the BPR
lifecycle.
Reusing Semantics and/or Syntax
A reference model can be used as a template in three ways. Quite simply, its
semantics and syntax can be re-used or only one or the other can be used.
Type of Re-use
Part for re-use True Template Semantic Foundation
Modelling training aid
Semantics X X Syntax X X
Figure 6: Types of re-use and parts of the model re-used
The first and second cases have been named True Template and Semantic
Foundation, and are examined next. The third type, termed Modelling
Training Aid, has only limited application for training on how to model using a
specific modelling technique and is not discussed further.
True Template
The first option, the true template, starts with the reference model and
directly edits it to produce the required enterprise specific model, using both
the syntax and semantics of the reference model as shown in Figure 7. To
use a reference model as a True Template it must have the “ability to edit”
characteristic of “open”.
Masters Thesis Chris Taylor
QUT Page 47
Figure 7: Using a Reference Model as a True Template
Semantic Foundation Only
Alternatively, the reference model can serve as a semantic foundation only,
providing or supplementing the designers’ knowledge about the domain. This
second semantic-only method would be required if the desired enterprise
specific model was in a different language to the reference model and would
also necessitate interpretation of the reference model as depicted in Figure 8.
Figure 8: Using a Reference Model as a Semantic Foundation
In either case, whether re-using the syntax and semantics or semantics only,
several derivation methods can be employed.
Derivation Methods
Four methods for derivation are described below.
Reference Model Enterprise Specific Model
Derivation
Red hexagon
Blue Circle
Green Square
Yellow Cross
Reference Model Enterprise Specific Model
Derivation
Masters Thesis Chris Taylor
QUT Page 48
Customisation
The first is “Customisation” graphically explained in Figure 9. This involves
changes to the model including the addition, deletion and/or modification of
certain parts of the reference model. This could involve changing objects,
names or layout of the model or adding completely new parts of the model.
Customisation is shown in Figure 9.
Customisation might be as simple as changing the name of the reference
model to reflect the organisation using it.
Figure 9: Derivation by Customisation
The above picture shows how the majority of the reference model is reused
with small changes.
Red-lining
A sub-set of the Customisation type of derivation is Red-lining. “Red-lining” a
term previously used to describe the process of deriving the enterprise
specific model from the SAP reference model e.g. Curran et al. (2000), is
used to describe the process where parts of the model are deleted, but not
added, during the derivation process. The rest of the model is used without
any modifications.
Reference Model Enterprise Specific Model
Derivation
Masters Thesis Chris Taylor
QUT Page 49
Figure 10: Derivation by Red-lining
Configuration
Figure 11 shows the Derivation by “Configuration”, this requires that the
reference model has explicit configuration points (i.e. its “Readiness for Use”
characteristic is “Contains Variants”). This is similar to the Red-lining
derivation, but in this case a reference model user is forced to make a
decision between two (or more) alternatives.
Figure 11: Derivation by Configuration
Figure 11 above shows how the reference model user selects between
explicit alternatives, in this case the red hexagon and not the blue circle, and
the aqua square and not the yellow cross.
Reference Model Enterprise Specific Model
Derivation
Reference Model Enterprise Specific Model
Derivation
OR
OR
Masters Thesis Chris Taylor
QUT Page 50
Embellishment
The third type of derivation is “Embellishment”. Figure 12 depicts the
Embellishment step, where extra detail is added to the reference model in
order to make it suitable for the enterprise specific model. Whether a model
requires embellishment is closely related to the characteristic of Level of
Abstraction of the reference model. The embellishment could also involve not
only adding higher level of detail but also other information such as different
views or performance, implementation or other data.
Figure 12: Derivation by Embellishment
In figure 12 the basis high level nature of the reference model has been
reuses, but the enterprise specific model has added detail at a lower level.
Many real life derivations will probably be a combination of the above
derivation methods, dependent on the characteristics of the reference
model/s that has/have been selected and the requirements of the enterprise
specific model. Generally the characteristics of the reference model
(specifically the “Readiness for Use” characteristic described in 2.3.4)
requires different combinations of the derivation methods as shown in Table
6. The “yes” and “no”s indicate which particular derivation method/s is/are
required for the type of reference model being used.
Although theoretically possible, the use of a reference model without a
degree of customisation is not considered here, because in practice it would
not occur. The customisation in the extreme case would require only the
Reference Model
Derivation
Enterprise Specific Model
Masters Thesis Chris Taylor
QUT Page 51
changing of the name of the reference model to suit the individual situation in
which it is used.
Derivation Method Required Customisation Red-lining Configuration Embellishment
Single depiction Yes Yes/No No No
Contains variants Yes Yes/No Yes No
Rea
dine
ss fo
r Use
Ty
pe
Abstract Yes Yes/No Yes/No Yes
Table 6: Readiness for Use Type and Type of Derivation Required
Having defined above what parts of the model can be used as a template
and linking this to the types of derivation, the next step looks at the individual
applications of the reference model as a template. A reference model can be
used as a template (in any of the ways described above) three times in the
process management lifecycle, once for the identification of processes, which
would use the higher level abstractions of the reference model, again when
modelling the more detailed AS-IS processes and thirdly for the development
of the TO-BE models.
Applications of Reference Model as a Template
The three applications as a template are highlighted in Figure 13.
Masters Thesis Chris Taylor
QUT Page 52
Figure 13: Applications as Template
Process Identification
Firstly, in the case of the higher level model reference model being used for
process identification, the goal is simply to provide a structure and scope for
the targeting phase. It provides a starting point which an organisation can
modify to suit its own terminology and process structure at a high level
(Huxley 2003 p92-3). Once this is complete the results can be used as a
comprehensive documentation of the processes in which an organisation is
operating, i.e. it has identified the processes. This is the input to the targeting
phase, which questions on which processes effort should be expended for
process improvement.
Rosemann’s original “Process Identification” step has been split into pure
identification, essentially the cataloguing of the organisation’s processes, and
process targeting, which takes this list of processes and selects those that
will go through the remainder of the lifecycle steps in a particular cycle. This
Select Reference Model
Process Identification High-Level Template
Process Targeting Scope definition/ suggestions
AS-IS Modelling AS-IS Template
Analysis Process Benchmark
TO-BE Modelling TO-BE Template
Process Implementation
Implementation information
Process Execution
Monitoring/Control Run Time
Information/Performance Benchmark
Masters Thesis Chris Taylor
QUT Page 53
is done to more accurately show the use of the reference model in these two
steps.
The reference models most suited for this identification and targeting stages
are the economic activity specific business process models at a “Complete”
Level of Detail, as they depict the high level processes that are common
across the industry being examined.
AS-IS Process Modelling
Again the model can be used as a template or a guide in the AS-IS modelling
phase. At this stage the mid-level models can be used to segment the
modelling work and act as a framework for the lower level detail. The lowest
level detail in the reference model can be used as a semantic check,
ensuring that modellers have a basic understanding of the process. This is
important as it allows the enterprise modellers, who may lack domain
knowledge, to be able to probe the process owners and executors to ensure
that the AS-IS model is accurate. It is proposed that this understanding of the
domain provided by the reference model can significantly improve the
developed models, particularly when the modellers are not experts in the
particular domain. A reference model that would be useful at this stage could
be either an Intermediate or Task level model of the specific economic
activity or of a specific functional area. Depending on the maturity of the
processes being examined a common practice (for immature organisations)
or best practice (for mature organisations) would be more appropriate as they
would more closely reflect the AS-IS situation.
TO-BE Process Modelling
Assuming that process improvement is one of the goals of the modelling
project and that the reference model chosen is of the type “best-practice”, the
reference model could be used as a template for the TO-BE model
development. If the model was ‘single depiction’ then limited work would
need to be done on the reference model to transform it into the TO-BE
model.
Masters Thesis Chris Taylor
QUT Page 54
2.6.2 Scope Definition and Targeting Suggestion
This section deals with the application of the reference model as a scoping
and or process targeting suggestion tool as shown in Figure 14.
Figure 14: Scope and Definition and Targeting Suggestions
Used in the Process Targeting Step, a reference model, either implicitly or
explicitly, suggests where an organisation should focus its process
improvement activities, as alluded to in Huxley (2003). He states that, when
using a reference model for process identification and targeting, processes
may not be recognised and that this lack of recognition could be due to the
low importance of the processes. (Huxley 2003 p.92-3).
The choice to include something as a core, support or strategic activity, the
naming or hierarchical positioning of a process in the model, can all implicitly
suggest the reference model creators’ perception of the importance of a
process.
Select Reference Model
Process Identification High-Level Template
Process Targeting Scope definition/ suggestions
AS-IS Modelling AS-IS Template
Analysis Process Benchmark
TO-BE Modelling TO-BE Template
Process Implementation
Implementation information
Process Execution
Monitoring/Control Run Time
Information/Performance Benchmark
Masters Thesis Chris Taylor
QUT Page 55
Additionally, the extended information included with a model might include
suggestions about the importance or criticality of processes. For example, a
study may suggest “critical” processes for a particular industry, or may
provide data on past improvements in a particular process in other
companies that could be extrapolated into the current situation to aid the cost
benefit analysis. A list of critical processes could be linked to strategic
directions, for example in a consumer market a particular vendor may choose
to be a high-end market niche player, the reference model may contain
information that suggests that, when using this strategy, marketing or quality
control processes are highly critical. The reference model could also
incorporate methodologies to determine the most critical processes for
example by showing cause and effect relationships between processes, or by
linking processes to strategy, goals or industry standard key performance
indicators (KPIs).
2.6.3 Process Benchmark
The reference model can be used as a process benchmark. Particularly if the
reference model was used a true template as described above, the
comparison of a reference model ‘Best Practice’ with the AS-IS models can
be relatively straight forward, allowing a gap analysis. The use of process
models for benchmarking can make assessment very specific and can
highlight opportunities for improvement that other methods may miss
(McGowan and Bohner 1993). This gap analysis can then aid the process
issue identification and therefore provide suggestions for improvement which
can be incorporated in the TO-BE modelling phase. This step would be most
beneficial if the reference model chosen was a ‘best practice’ model,
although the comparison of the AS-IS to a ‘common’ practice reference
model would still highlight potential areas for analysis, particularly if
standardisation was the key driver for the business process re-engineering.
2.6.4 Implementation Guide
As can be seen in the classification scheme, some reference models include
information on how to implement the proposed processes, including
information such as change management information, previous experiences
Masters Thesis Chris Taylor
QUT Page 56
and lessons learned, or models such as procedural models with steps for
implementation.
2.6.5 Run Time Information
Models that contain run-time information such as including for example
advice on staffing, technologies, reoccurring problems and methods for
overcoming them, can be used in the Process Execution phase of the
Business Process Lifecycle.
2.6.6 Performance Benchmark and Template
More comprehensive and ‘extended’ reference models can suggest relevant
performance metrics and even provide comparative data that may have been
collected in other instances. This Key Performance Indicator (KPI) template
can then be modified to suit the individual project to assist the identification of
performance measures, and then at run-time it can assist with the monitoring
and controlling phase of the BPM lifecycle.
2.7 IT Service Provision
As described in Chapter 1, process modelling has been used in the
application areas of IT as part of the analysis of key business processes
however has had only limited use in IT service provision.
Those models that do exist, focus on the early stages of IS and IT systems
lifecycle, for example in the selection or design, and implementation steps
(Gulla and Brasethvik 2000). Examples of models that suit these purposes
are the software spiral model, the SAP R/3 Implementation Guide, ValueSAP
or the Rational Unified Process and many implementation methodologies
from the consulting firms.
An area which is underdeveloped in terms of methodological support is the
administration, maintenance and support of the system. As the large number
of systems implemented in the late 1990’s mature, the support and benefits
realisation of these systems has become more important. Indeed in terms of
lifecycle costs, Gartner research shows up to 80% (Anonymous 2003) is
incurred the after the purchase of an IT system. The importance of
Masters Thesis Chris Taylor
QUT Page 57
maximising the benefits and minimising the costs of such IT investments is a
major motivator for this research. Not only is the efficiency and cost of the
provision itself important, but the effectiveness of the delivery can have
profound effects on the business processes it supports. A poignant reminder
of organisation’s reliance on IT is highlighted by the recent collapse of the
Australian Telco One.Tel due to a fault in its billing applications (Barry 2002).
This section of this literature review looks at the specific domain of the IT
service provider. It also examines some of the existing models for IT service
provision. This thesis focuses on outsourcing of IT, and organisations whose
core business is IT service provision. IT outsourcing has been defined as
passing IT functions previously performed in-house to external parties (Gupta
and A 1992). At the turn of the 21st century the growth rate for IT outsourcing
worldwide was around US$100 billion and growing at around 16% p.a.
(Lacity and Willcocks 2000). A recent IDC report indicates that in Australia
alone the IT Outsourcing industry is worth over $AU10b (Benson 2002).
Sriram quoted Garter Dataquest (July 2002) as predicting that the Australian
IT services industry will grow up to $US16.7b by 2005 (Sriram Year
Unknown).
The focus on the IT outsourcing arrangement as the business model of
delivering IT services was chosen because of the increasing popularity of IT
outsourcing and the expectation that IT outsourcers perform better than in-
house service providers, and provide an opportunity to clarify and enforce the
terms of the service as well as penalties for breaches (Lacity and Willcocks
2000).
This focus, however, does not mean that the findings or the reference model
would be entirely unsuitable for internal service provision. Indeed it has been
argued that internal service providers should, to a certain extent, model
themselves on external providers to ensure efficiency, accountability and
effectiveness (Ross et al. 1999).
Most of the literature on IT services focuses on outsourced IT services. It has
been defined as:
Masters Thesis Chris Taylor
QUT Page 58
“the contracting by an organisation with a third party for the
management and enhancement of ongoing operations for all or part of
its IT infrastructure, IT functions, business processes, or business
solutions. Outsourcing involves a fixed-term, contractual arrangement
that may involve the transfer of assets or people.” (Benson 2002)
The actual services that constitute IT Services are show in Figure 15.
Figure 15: IT services as defined by Benson (2002)
As noted in the same report IS Outsourcing makes up 38% of the IT services
market. The range of outsourced IT services providers spans from the most
basic such as network service providers, to the more business-oriented types
such as business service providers. The following examination of the IT
services market allows the reference model to be positioned.
In order to present a taxonomy of IT outsourcing in which to position the
reference model, the Application Service Provision (ASP) market, a subset of
the IT services market is examined in the next section. The ASP market
layering is extrapolated to the IT services domain because much literature
has been produced about the ASP, and how the market is segmented.
2.7.1 Application Service Providers
In the mid 1990’s the concept of Application Service Provision (ASP) was
introduced. It is a form of applications outsourcing. Much literature
particularly industry press was devoted to explaining how ASP was
positioned in the market.
Although the idea is simple, “ASP” however is a somewhat confusing term
(Dunn 2001; Swift 2001). The definition and scope of available ASP services
continues to evolve, muddying the term (Boyle 2002).
Masters Thesis Chris Taylor
QUT Page 59
The idea behind Application Service Provision (ASP) is not new. ASP is a
type of IT outsourcing. As early as the 1960’s traditional bureau service has
offered companies access to payroll processing (Bennett and Timbrell 2000;
Perdue 2000; Swift 2001). Other examples of outsourcing include the “time-
sharing” concept in the mainframe days of early computing (Perdue 2000;
Swift 2001).
Some authors specify that ASPs deploy, host and manage “packaged”
(Zimmerman 2002) or “off-the-shelf” (Bennett and Timbrell 2000)
applications, while others include in ASP those who develop or host specially
developed or unique software (Swift 2001). Other ASP services can include:
implementation, configuration, maintenance, support and application hosting
(Bennett and Timbrell 2000).
Other authors specify that the ASP physically host the hardware and client
data (Swift 2001; Anonymous 2002) which is essentially Application Hosting.
ASP is sometimes defined as the one-to-many business model, leveraging
the infrastructure to service several clients from the same capital base (Swift
2001; Boyle 2002). While others contend that a one-to-one relationship does
not preclude an organisation from being seen as an ASP (Lavery 2001).
Differing definitions of the ASP would include services such as web-based
email (e.g. hotmail, yahoo); however other definitions would exclude
applications that are not available ‘off-the-shelf’ (Timbrell et al. 2001).
Summarising the literature, the term Application Service Provision refers to
the specific business model of leveraging investments in a single back-office
infrastructure (i.e. hardware, database and software) to deliver application
functionality over a network to multiple clients paying rental or other usage
based fees. The ASP is responsible for every part of the delivery up to the
client’s local network (and possibly including a customer-side thin client). It is
a form of outsourcing; however it differs from traditional outsourcing
arrangements in 5 significant facets summarised in Table 7 (adapted from
Yao and Murphy 2001).
Masters Thesis Chris Taylor
QUT Page 60
Table 7: Traditional Outsourcing vs. ASP (Yaho and Murphy 2001)
A good example of the industry stratification is presented in Figure 16
(Seymour and Edwards 2001).
Figure 16: xSP Taxonomy (from Seymore and Edwards 2001)
Although the taxonomy presented specifically depicts only the ASP model of
delivery, it serves as a good basis for IT services as a whole. ASP is a
specific delivery method of IT Services; however the layers are still valid at a
generic level. Another layering of the ASP market presented in Figure 17,
also provides a basis for the layering of the IT services market as a whole
(Tao 2001).
Figure 17: Four layered stratification of the ASP market (from Tao 2001)
Masters Thesis Chris Taylor
QUT Page 61
2.7.2 Taxonomy of IT Services
ASP is a specific delivery method of IT services, it does however provide a
good overview of the layers needed to deliver IT services regardless of the
method. Generalised from the xSP taxonomy and from Tao’s 4 layers one
can derive an IT service stratification presented in Figure 18. Placed beside
each of the layers is each service defined in Figure 15 which defined the
separates components of IT services. The IT industry is an extremely
complex environment, and therefore any simple classification scheme should
be seen only as a heuristic.
Figure 18: Simplified Taxonomy of IT Services
IT services can be supplied by a single (usually large) organisation although
the supply chain is typically filled with niche players. For example, company
A could provide the technical infrastructure and hosting while company B is in
charge of technical maintenance and application of the system, and company
C provides database support and tuning while company D provides support
services to end-users, while company E manages the Information
Communication Technologies (ICT).
A brief discussion of the layers presented above is presented here for clarity.
Solution Delivery
Service Delivery
Infrastructure Management
Network Management
Consulting, Education, Business Processing Services
Software Support, Hardware support Configuration Management,
Capacity Management
Application Hosting, Data Centres, storage centres
Network Management
Masters Thesis Chris Taylor
QUT Page 62
Network Management
In this case the network is the transportation networks which connect the IT
infrastructure to the local network of the end user, assuming they are
geographically separated. Hence, network management is the planning,
organising and control of this network. It involves tasks such as managing
negotiations with other network providers, ensuring quality of service at the
network layer, ensuring connectivity, managing network resources such as
cabling and switching stations. Network management is seen as a subset of
the activities provided by the Information and Communications Technology
Industry (ICT).
Infrastructure Management
Infrastructure management is what some would describe as “looking after the
boxes”. Infrastructure in this case refers to the infrastructure from which the
IT services are being provided excluding the transportation networks. It
includes the internal network of the end user organisation and the network of
the infrastructure centre, the actual physical infrastructure for the housing of
the IT equipment, particularly the servers and hard-drives. The scope
includes providing services such as UPS, physical security, certain disaster
recovery mechanisms such as redundant or lights-out facilities. This layer
captures many of the processes associated with application hosting.
Service Delivery and Support
Service support involves dealing with problems in the IT environment. It is
largely re-active though it can have proactive elements. It involves activities
such as receiving incident reports from users or others, finding solutions to
these problems and implementing them in the production systems. Service
delivery is concerned with all the ongoing activities that ensure the continued
operation of the system for the end user. It includes activities such as
configuration management, capacity management and service level
management. This layer includes the administration of applications including
ERP’s.
Masters Thesis Chris Taylor
QUT Page 63
Solution Provision
Solution provision is a wide catch-all for the ‘value-add’ or extra activities that
can be provided to supplement the IT services. Examples include wrapping
the underlying layers into a single package and providing a single face for the
customer, providing consultancy on which IT service provider and packages
to choose, or even business process re-engineering, business process
outsourcing or strategic IT guidance.
2.7.3 IT Services Lifecycle
IT Services also follow a relatively standard lifecycle. Presented here are
several lifecycle models that have been used to derive the IT services
lifecycle used in this thesis.
The Microsoft Operating Framework (MoF) documents best practices,
principles and models for operating an IT infrastructure. The MoF consists of
4 phases “Planning”, “Preparing”, “Building” and “Operating”.
CORBIT, described in more detail in 2.8.2, (Guldentops et al. 2000) contains
a high level lifecycle of IT services. The 4 major phases are “Planning and
Organisation”, “Acquisition and Implementation”, “Delivery and Support” and
“Monitoring”. The following definitions are quoted directly from the CORBIT
framework:
“Planning and Organisation
This domain covers strategy and tactics, and concerns the
identification of the way IT can best contribute to the achievement of
the business objectives. Furthermore, the realisation of the strategic
vision needs to be planned, communicated and managed for different
perspectives. Finally, a proper organisation as well as technological
infrastructure must be put in place.
Acquisition and Implementation
To realise the IT strategy, IT solutions need to be identified, developed
or acquired, as well as implemented and integrated into the business
Masters Thesis Chris Taylor
QUT Page 64
process. In addition, changes in and maintenance of existing systems
are covered by this domain to make sure that the life cycle is
continued for these systems.
Delivery and Support
This domain is concerned with the actual delivery of required services,
which range from traditional operations over security and continuity
aspects to training. In order to deliver services, the necessary support
processes must be set up. This domain includes the actual processing
of data by application systems, often classified under application
controls.
Monitoring
All IT processes need to be regularly assessed over time for their
quality and compliance with control requirements. This domain thus
addresses management’s oversight of the organisation’s control
process and independent assurance provided by internal and external
audit or obtained from alternative sources.” (Guldentops et al. 2000)
p16.
The lifecycle model and terminology used in this thesis re-uses these
concepts and is presented in Figure 19. It defines the 4 phases of planning,
implementing, support and delivery and retirement.
Figure 19: IT Services Lifecycle
A brief description of each of the stages is presented next. This lifecycle
tracks the relationship between the provider and consumer, and it could have
several iterations particularly of the planning and implementing stages.
Plan
Implement
Support
Retire
Masters Thesis Chris Taylor
QUT Page 65
Planning
This stage starts after an initial contact between the service provider and
consumer has been made. It involves the matching the needs of the
consumer with the provider’s services, and if necessary a degree of
customisation of the services on offer. Assuming the consumer is satisfied at
the end of this stage, the implementation or go-live can commence.
Implementation
This involves the roll-out of the service and any required changes.
Transitioning of any legacy systems, human or other legacy resources also
happens in this stage.
Delivery and Support
In this stage, the actual run-time value add service takes place. This phase is
concerned with providing and maintaining the agreed levels of service, in
order to enable the processes of the consumer.
Retirement
Should the services of the particular service provider no longer be required,
then the retirement is the phase that disengages the provider from the
consumer. This could include handing over data, cancelling legal obligations,
possible re-allocation of staff and eventually decommissioning the service.
Often the retirement of a service may be integrated with the planning of a
new service designed to support the same needs.
2.8 ITSP Business Reference Models
The previous section explored IT services and positioned the model in terms
of IT services market layer and lifecycle phase. As introduced in Chapter 1,
proven business process management techniques have not been applied to
the IT service provider domain.
This section is in two parts, firstly, a description of the reference model that
could help to apply the business process management techniques to the IT
service provider domain, then an examination of the current related models.
Masters Thesis Chris Taylor
QUT Page 66
2.8.1 Desired IT Service Provider BP Reference Model
From the discussion in Chapter 1 and the linkages to BPM described in
Chapter 2 we can derive the requirements of a suitable reference model. It
would obviously describe the business processes of an IT service provider.
The model would cover the complete scope of the IT service provider
organisation but also contain a single, detailed and integrated model,
containing accepted “Best Practice” processes.
The model should be specifically for those that specialise in IT services, i.e.
the IT service provider organisations, the IT outsourcing vendors, be updated
as practices change, and to be used as a true template to save derivation
costs.
Modern BPM techniques require the use of semi-formal modelling languages,
to ensure the consistency of the model and the language used in the model
and to help the translation into IT based support solutions.
Masters Thesis Chris Taylor
QUT Page 67
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 8: Desired characteristics for IT Service Provider BP Reference Model
2.8.2 Existing Models related to IT Service Provision
This section looks at selected Business Process Reference Models for IT
service provision that have served as input to the proposed reference model.
A limitation of this section is the access to proprietary knowledge in the form
of IT Service Management (ITSM) reference models, specifically those held
by ITSM and related activities vendors, including consultancy companies and
IT Management software and services companies. This thesis excludes
those models to which the research team could not obtain access.
Masters Thesis Chris Taylor
QUT Page 68
ITIL (Information Technology Infrastructure Library)
ITIL, which had its genesis in the 1980’s, is a publicly available best practice
definition of IT service processes originally produced by an arm of the UK
government. ITIL concentrates on IT service management (ITSM) and is a
living model. One of the major contributors to the continuing development is
an independent industry group, the IT Service Management Forum (ITSMf at
www.itsmf.com). Since its early introduction ITIL has been updated and
refined with input received via various channels and a large and active user
group particularly in Europe. ITIL is a widely accepted best practice model
and contains detailed process descriptions at a Task level. However, ITIL’s
scope is tightly focused on ITSM to the exclusion of other IT service provision
(ITSP) processes. For example, supplier interface, HR, CRM and other
processes are not covered in ITIL. Another limitation of the ITIL models is the
lack of top down structure. Possibly due to the fractured nature of the
creation of ITIL, it was developed originally as around 60 individual booklets
on isolated areas of IT, with different authors and perspectives, it is
frequently inconsistent and the processes are difficult to fit together and
consolidate.
An attempt to provide a high-level integration of the ITIL process models has
been produced by Enterprise Managed Services in the EMS Process
Reference Model (Enterprise Managed Services, Unknown) shown in Figure
20.
Masters Thesis Chris Taylor
QUT Page 69
Figure 20: EMS’s High level ITIL process model
ITIL also lacks integration of its process elements as acknowledged by the
following quote from the ITIL documentation:
“The major elements of the ITIL books can be likened to overlapping jigsaw
puzzle pieces … , some of which have a precise fit, and some of which
overlap or do not fit together accurately.” (CCTA 2000 section 1.4).
ITIL is also inconsistent in presentation, as shown in Figure 21 which comes
from (left) Annex 5D and (right) chapter 6.6.
Business/Customers
IT - Business Alignment Service Support
New
Projects
Business
requirements
Service
Definition
CRM Service
Desk Incident
M’ment
Problem
M’ment
Release
M’ment
Config
M’ment
Change
M’ment
Cost
M’ment
Contigency
M’ment
Capacity
M’ment
Avail.
M’ment
SLM
Storage
M’ment
Desktop
M’ment
App. and
DB M’ment
Systems
M’ment
Network
M’ment
IT Environment
Service
Delivery
Infra.
M’ment
Masters Thesis Chris Taylor
QUT Page 70
Figure 21: Examples of inconsistencies in modelling in ITIL documents
The classification table of ITIL is presented in Table 9.
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 9: ITIL Classification
Masters Thesis Chris Taylor
QUT Page 71
IT Service Management Capacity Maturity Model
The Information Technology Service Capacity Maturity Model (IT Service
CMM), is a result of several projects involving universities, private
organisations and government departments based mainly in Europe. The
model attempts to define what functions should be in place to reach a certain
level of maturity. It is a form of quality management because it provides the
what, but not the how, i.e. the functions but not the processes.
“The IT Service CMM only specifies what practices are needed, not how they
need to be implemented” (Niessink et al. 2002 p12)
Its objectives are two fold, to allow an assessment of the current state of an
organisations IT service provision as well as providing directions and steps
for improvement of the service provision. The model references ITIL as a
good source of the “how” these processes should be conducted.
Figure 22: IT Service CMM Methodology
Masters Thesis Chris Taylor
QUT Page 72
The actual model is built using the methodology outlined in Figure 22. Along
with the definition of Key Process Areas for a particular maturity level, the
model defines the goals of each of these areas and also the common
features. The common features describe mostly organisational and policy
issues that commonly exist to support the process area.
Each common feature is then examined using a lifecycle-like model
(Commitment to Perform, Ability to Perform, Activities Performed,
Measurement and Analysis, and an overarching Verifying Implementation),
and key practices within each of these lifecycle steps is provided.
The high level view of the model showing the levels of maturity with the
associated processes is provided in Figure 23.
Figure 23: The IT Service CMM
Masters Thesis Chris Taylor
QUT Page 73
The classification of the IT Service CMM is presented in Table 10.
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 10: IT Service CMM Classification
MoF (Microsoft Operating Framework)
The MoF, developed by Microsoft, references several ITIL definitions in its
attempt to provide the technical guidance within the Microsoft “Enterprise
Services” (Microsoft Corporation 2001). The Enterprise Services outlines the
lifecycle in planning, preparing, building and operating lifecycle for Microsoft
technical solutions. The MOF acknowledges ITIL as the current industry best
practice for IT service management. The MOF only uses selected parts of
ITIL and offers no additional information in terms of ITSM or ITSP processes.
It is focused on technical issues. Also, in its attempt to model the lifecycle of
system implementation, its focus is not on providing ITSP processes.
Masters Thesis Chris Taylor
QUT Page 74
Figure 24: Microsoft Operating Framework
The overview of the scope of MoF is presented in Figure 24. The
classification is presented below.
Masters Thesis Chris Taylor
QUT Page 75
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 11: IT Service CMM Classification
Garschhammer et al.
Garschhammer et al.’s work on building a service provider model
acknowledges the increasing importance of the supply chain in service
provision (Garschhammer et al. 2001). Using a top-down methodology
Garschhammer et al. have produced a generic service model that defines
commonly needed service related terms, concepts and structuring rules.
Garschhammer et al.’s top-down modelling approach creates the opportunity
to capture all SP processes. It is however, lacking finer granularity and fails
to provide a structure for further refinement into process models.
Masters Thesis Chris Taylor
QUT Page 76
Figure 25: Garschhammer et al. Service Model
The classification is presented below.
Table 12: Garschhammer et al. Classification
Masters Thesis Chris Taylor
QUT Page 77
TOM (Telecom Operations Map)
The TOM was developed to define the business process for the Information
and Communication Services Industries. Introduced in 1995 under a different
name, the TOM was developed by the TeleManagement Forum. It provides
guidance on systems delivery, system management and billing. TOM
focused exclusively on the processes directly related to service provision, to
the exclusion of supporting processes. TOM has since been superseded by
eTOM (see below).
eTOM (expanded Telecom Operations Map)
eTOM is an extension of the TOM and provides a business process
framework for service providers (TeleManagement Forum 2001). eTOM
provides a top down hierarchical structure into which processes, or at least
functions, can be classified. As well as providing this framework, eTOM gives
an indication of the interactions and dependencies of these functions on one.
Although eTOM is aimed at the Information and Communication Services
industry at higher levels, the generality of the model makes it applicable for
other service providers in the IT and related industries. This model attempts
to show the interrelations between identified processes, but is sometimes
confused between linking events and data dependencies. This lack of
consistent sequential and temporal order of activities makes eTOM less of a
process model and more of a classification framework for processes or
functions with some interrelationships between the processes shown at the
lower levels.
Figure 26 shows the highest level of the eTOM model (Anonymous 2001
p.20).
Masters Thesis Chris Taylor
QUT Page 78
Figure 26: eTOM Business Process Framework - Level 0 Processes
eTOM’s classification is presented below.
Masters Thesis Chris Taylor
QUT Page 79
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 13: eTOM v2.5 Classification
CORBIT
Control Objectives for Information and related Technology (CORBIT) is
positioned as a model for IT Governance, which it defines as
“A structure of relationships and processes to direct and control the
enterprise in order to achieve the enterprise’s goals by adding value
while balancing risk versus return over IT and its processes.”
(Guldentops et al. 2000 p.5)
It is aimed at providing the monitoring guidance to ensure that business
needs are met by IT operations and conforms to seven principles (e.g.
effectiveness, availability). In effect it is a large IT management performance
checklist, providing information on what aspects of performance to measure
Masters Thesis Chris Taylor
QUT Page 80
and recommendations on how to measure them. It also provides a high-level
breakdown of the functions involved with the four defined lifecycle steps of IT
services (see section 2.7.3 for a discussion of the 4 lifecycle steps).
Figure 27: Corbit Overview
The overview of the CORBIT documents is presented in Figure 27. Although
Corbit provides a good basis for the control and auditing of IT management it
is not a process model. Its classification is presented below.
Masters Thesis Chris Taylor
QUT Page 81
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 14: CORBIT classification
2.8.3 Perceived Shortfalls in existing ITSP BPRM
IT service provision is a complex task, involving detailed technical issues,
people and management issues, organisational issues, frequent changes
and a high degree of customer interaction. It is also highly dependent on the
context of the process. Factors affecting the IT service provision include the
types of services provided, the relationship between the service provider and
customer or the size of either organisation along with many others.
Hence any attempt to provide a business process model for this domain is a
complex undertaking. The high degree of interaction and dependencies of
the processes in an IT service provision scenario also make the task of
mapping them highly complex. A model of such a situation needs to be
Masters Thesis Chris Taylor
QUT Page 82
detailed enough so it is helpful in a practical setting, and yet abstract enough
so that at a higher conceptual level the interactions and dependencies can be
easily understood. The consistent and integrated linking of the higher-levels
with the lower levels is also extremely important, because one relies on the
other. The higher-levels depend on the lower-levels for the detail to make the
model useful and the lower-levels depend on the higher to provide a
conceptual entry point for the model, and to facilitate communication about
the domain. Most of the models lack this detailed level of descriptions and
are often only functional representations, not true process definitions.
ITIL contains the most detailed process information (it is the only one rated in
the Task level of detail other than the MoF which references ITIL heavily) and
its popularity in industry (Morin 1999; Duffy 2001; Dubie 2002; Kara 2002)
could partly be attributed to this level of detail, which gives managers and
operators tangible grass-roots ideas and suggestions. The other models are
higher level and lack this detail needed to make a difference at the
operational and implementation level.
ITIL lacks this higher level. It contains inconsistencies in content and format
and is very hard to conceptualise at the higher levels. Taken in isolation, the
advice it offers is useful, but stepping back from this detailed view and trying
to fit the pieces of information together is a difficult task. ITIL is also not
supported in a modelling tool and, although it claims to be applicable in the
commercial IT service provider domain, it is not specific to this environment,
and indeed is more focused on in-house service provision.
It is the aim of this research to fill this gap by taking parts of the content and
structure of these models to provide a reference model that clearly shows the
higher level dependencies and interactions while at the same time providing
clearly defined and detailed guidance on the business processes at the Task
level for IT service provision.
2.9 Chapter Summary
This chapter has provided a review of the literature on reference models from
which a classification scheme for reference models has been derived. It
Masters Thesis Chris Taylor
QUT Page 83
reviewed reference models related to IT Service Provision and presented a
brief discussion of IT service provision. From this discussion an IT services
market layering and lifecycle was derived. As a culmination of the literature
review the perceived weaknesses of the currently available business process
reference models for IT service provision have been identified. ITIL is the
only identified source that contains detailed advice about IT Service provision
at the Task level. The major weakness of ITIL however is its lack of top down
integration and consistency. The next section describes the research
methodologies applied to reach the stated goal of a reference model for IT
service provision.
Masters Thesis Chris Taylor
QUT Page 84
Chapter 3: Research Design and Method
3.1 Introduction
The main goal for this research is to produce a partially completed reference
model for IT service provision. This goal produces several research
questions as outlined in Chapter 1. Due to the lack of work in the area of
reference models, their characteristics, development and use methods,
exploratory research methods are required to examine the identify the
concepts in the field. This chapter outlines the research design and
methodologies embedded in the design, which were employed to accomplish
the research goal and answer the research questions.
Firstly, the Research Design is presented, followed by sections on the Focus
Group research methodology and secondly the case study research
methodology. This chapter describes the research methods in a general
sense and provides justification for their use in this particular research
program. The specifics of the research methodologies as they are applied in
this research are presented later in the thesis, before the results of the
research stages are presented. This is done to allow the reader to examine
the results of the research stage with respect to its individual setting.
3.2 Research Design
Keeping the research goal in mind, that is the design of a reference model,
the research was designed in 3 major stages. The first stage, involved a
comprehensive literature review the results of which have been presented in
Chapter 2. The aim of this stage was to provide a basic understanding of the
domain of IT services and the concept of reference models. The second
phase of the research was the design of the models. It involved several focus
groups to validate and modify the models put forward by the author. The
outputs from this design phase were to be tested in case studies and the
opinions of those using the reference model examined.
During the literature review it became apparent that there was no literature
describing what aspects were important or desirable in a business process
reference model. This prompted research into the quality of business process
Masters Thesis Chris Taylor
QUT Page 85
reference models using the case study methodology which is presented in
Chapter 4. The experiences and early results gained through the case study
on model quality significantly influenced the reference model design.
Another deficiency in the literature was a description of how reference
models should be designed; the literature review did not reveal any detailed
guidance on how to design reference models. The very few articles that
involved the development of a reference model did not describe in any detail
the steps that were used in the design, or comment on how future reference
models should be designed. The design procedure used for this research is
presented in Chapter 5. The actual design and validation of the model are
described in Chapter 6 and the testing of the model in Chapter 7.
Figure 28: Research Design and embedded Methodologies
Each of the research methodologies used in this research are now described
in a general manner and their use justified. The specifics of the uses of the
methodologies, e.g. focus group, data collection and analysis, are presented
in the relevant sections of the thesis.
Literature Review
RM Quality Case study
protocol
Draft Model
BP RM Quality Attributes
Case Studies
Focus Groups
Validated Model
Case Studies
Tested Model
Masters Thesis Chris Taylor
QUT Page 86
3.3 Focus Group
3.3.1 Introduction
Focus groups were first used during the second world war, when government
officials desired to assess the effectiveness of their propaganda and soon
moved to explore the effect of films and television (Merton R 1956). Their
application has been extended to marketing and corporate image campaigns.
3.3.2 Characteristics
The focus group is an effective method for gathering the general opinion of a
target audience by providing an environment that allows probing for
clarification and justification of opinion (Morgan 1988 in Saulnier 2000a),
capture of peoples’ reactions to others comments (Saulnier 2000b), a degree
of spontaneity (Magill 1993 in Saulnier 2000a) and a degree of completeness
(Kruger 1994 in Saulnier 2000a).
Research into areas that have not been sufficiently explored require initial
exploratory work (Sofaer et al. 2001). Morgan (1988) suggests that focus
groups are a suitable research method for orienting a researcher to a new
field.
Morgan states that the "hallmark of focus groups is the explicit use of the
group interaction to produce data and insights that would be less accessible
without the interaction found in a group" (Morgan 1988 p.12).
3.3.3 Conducting Focus Groups
Focus groups consist of a ‘round-table’ meeting with researchers and
industry peers for the discussion of a precisely defined topic under strict
moderation. Focus groups generally have between 6 and 10 participants.
There are many examples of larger groups (Saulnier 2000b), although larger
groups are more difficult to manage and tend to allow some participants to
say nothing or follow the more dominant personalities (O'Neill et al. 1999).
Focus groups are typically 1 to 2 hours in length and involve three facilitators;
the moderator, note taker and time keeper (O'Neill et al. 1999).
Masters Thesis Chris Taylor
QUT Page 87
3.3.4 Justification
There were three reasons focus groups were used for the design of
reference models; the first historical, the second theoretical and the third
practical.
Firstly, from a historical perspective, existing reference models were largely
designed from group collaborations. For example, eTOM (version 2.5) has
listed amongst its authors some 49 individuals and over 30 organisations. It
is reasonable to conclude that the quality and acceptance of a reference
model depends, to some extent, on the number and quality of the participants
involved in the design process. Given such a participant pool the next
consideration is how to consolidate opinions to into the reference model. A
focus group as discussed above provides an opportunity to reflect and
compare on the opinions of others and as such is appears a suitable method
for the design of a reference model.
Secondly, from a theoretical perspective, the use of group discussions has a
backing for use in developing generalisable models as described by Frank
(Frank 1999). He argues that the only way to overcome subjective
perspectives while creating reference models is “the idea of a rational
discourse”. He goes on to argue that participants in this discourse should
constitute those that have “sufficient knowledge” about the subject, and that
ideally this should include everybody who is “directly affected by the artefact
under consideration”. For the purposes of this research the input from
everyone affected was impractical. Taking into account the need for a
consolidated view of the IT service provision domain, the focus group method
was chosen with representatives from large IT service providers (for further
details about recruitment see Chapter 3).
The third reason that focus groups were chosen for the model design was
that process modelling is usually done with modellers and several domain
experts (McGowan and Bohner 1993), effectively a small focus group. No
literature exists on how to create a reference model, so techniques have
been borrowed from the process modelling domain.
Masters Thesis Chris Taylor
QUT Page 88
3.4 Case Study
3.4.1 Introduction
The most commonly cited work describing the case study methodology is
Yin’s book “Case Study Research” (Yin 1994). Commonly cited works in the
Information Systems field include Benbasat et al. (1987) and Lee (1989).
The term case has been used in several contexts. It has been used to
describe a unit of analysis or to describe a research method (Myers 1997). It
is used in this thesis as the research method.
Myers (1997) quotes Orlikowski and Baroudi (1991) and Alavi and Carlson
(1992) as saying that case studies are the most commonly used qualitative
methods in information systems research.
Historically, case studies have been seen as an unscientific research
method, despite having been used in a variety of research areas including
social research, history and economics. Their widespread use, however is an
indication that this historical reluctance to accept this research method is
diminishing.
3.4.2 Characteristics
“The case study method refers to a group of methods which emphasize
qualitative analysis” (Gable 1991 p 3.1). It is defined as an “Empirical inquiry
that investigates a contemporary phenomenon within its real-life context” (Yin
1994 p.13) and can be conducted for exploratory, explanatory or descriptive
purposes (Yin 1994; Tellis 1997). Case studies are applied to serve an
exploratory function in this research to investigate a contemporary
phenomenon within its real-life context.
Yin (1994) recommends the use of multiple case studies, when the intent of
the researcher is to build and test a theory (Gable 1994; Yin 1994). Yin
suggests that a single case study is a relevant approach to discovering new
or un-researched material in a natural setting (Yin 1994).
Masters Thesis Chris Taylor
QUT Page 89
Benbasat, Goldstein and Mead (1987) add that case study “is an appropriate
way to research a previously little-studied area” (Benbasat et al. 1987).
According to Yin, there are at least 5 applications for case studies. They are;
to explain the effect of some intervention, to describe an intervention itself, to
illustrate certain topics, exploration of the effects of an intervention when the
outcomes of the intervention may not be clear, and in the meta-evaluation
(Yin 1994 p.15).
They are used in this research to illustrate certain topics, namely the user
perceived quality attributes of reference models, and to explore the effects of
an intervention, when presenting the developed reference model for use in
BPM projects and examining the perceived impact.
3.4.3 Conducting Case Studies
A comprehensive case study protocol should be derived to direct the case
studies by carefully documenting all procedures relating to the data collection
and analysis phases of the study. The protocol defines the structure of the
overall case study effort and is specially advantageous for exploratory
studies as this, for (1) they force the researcher to consider in advance, the
objectives and goals of the study, (2) to help avoid redundant effort, and any
potential omissions of the data collection and finally (3) to support the
communication and documentation efforts (Gable 1991; Yin 1994).
The sampling frame, (defining the contextual elements such as who to
contact, where to contact, how to contact etc.) should be derived based on
past theories and the study objectives. Qualitative data collection mechanism
such as in-depth interviews, observations, and content analysis of existing
documentation were conducted to collect ‘rich’ information about the
application of reference models within different contexts. All relevant data
should be maintained in a ‘case database’ (Miles and Huberman 1984; Yin
1994) and close linkage between the data, evidence and the research goals
must be maintained throughout the analysis.
Masters Thesis Chris Taylor
QUT Page 90
3.4.4 Justification
Case studies have been chosen as the research method for three reasons.
Firstly, the practical research into model quality and especially reference
model quality is extremely limited. Hence using an exploratory method, such
as case study, is appropriate. As shown in Figure 28, which describes the
research design, the case study method has been used twice. The first,
exploring the quality of reference models, is of Yin’s illustrative type. The
second, the examination of the use of the developed reference models,
explains the effects of the use of the reference models.
The case study offers an opportunity to observe real life behaviour and
provides a natural context, with all the pressures and interferences that that
entails. This was important for the research team, as this research is aimed
at the examining the use of reference models and their quality from the
perspective of those using them and as such should consider real life
constrains and impacts.
Thirdly, the feasibility of the case study method made it an attractive
opportunity. The research team was fortunate to have close access to
several organisations which were experienced in the area and which were
willing to provide their thoughts and time to this research.
3.5 Chapter Summary
This chapter has described the research methods used in this research and
provided justification for their selection for use as well as outlining the way to
conduct research using these methods. Further discussion of the specifics of
the research methods, e.g. participant selection is discussed before the
presentation of the results of the research.
Masters Thesis Chris Taylor
QUT Page 91
Chapter 4: Reference Model Quality 4
4.1 Introduction
During the review of the literature, very few publications were found
discussing exactly what a reference model should look like and contain. To
provide guidance and understanding for the development of the reference
model produced in this research, work was conducted to examine what were
the quality aspects of business process reference models, and what should
be included to supplement these models. Figure 29 shows how this chapter
fits in relation to the reference model lifecycle.
Figure 29: Chapter 4 in relation to the Reference Model Lifecycle
The research team wanted the derivation of the quality aspects of reference
models to be based on real life experiences, not only theoretical constructs,
and hence case studies were conducted.
This chapter details these case studies, their motivation, the terminology
used throughout the work, the design, data collection, analysis, results and
conclusions.
4 Parts of this chapter is published in the proceedings of the Australian Conference on Information Systems conference in Perth, November 2003 in the paper Taylor and Sedera “Defining the Quality Attributes of Business Process Reference Material”.
Design Reference
Model
Reference
Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 92
4.1.1 Motivation
The motivation behind this study was to understand what defined reference
model quality in order to incorporate those aspects which were important to
quality into the reference model.
4.2 Terminology
To clarify the discussion this section introduces the terminology used to
describe quality in this paper.
A brief review of the marketplace for reference models suggests that
producers, particularly commercial producers, are offering not only the
reference model but also a range of other services and products related to
the reference model (e.g. training, case studies, feedback mechanisms)
which is termed model support. The overall package offered by the producer,
i.e. the model and the model support, is termed the reference material. Of
course the major part of the reference material is the reference model and
hence is the focus of our research.
The Quality Framework includes the Quality Dimensions, which are high level
constructs or the goals of quality. The quality dimensions are at a conceptual
level and serve as a categorisation for the quality attributes. Quality attributes
are the sub constructs of the dimensions. They are the practical means of
achieving these high level quality dimensions as described in Figure 30. They
can also be seen as the ends, because achieving these attributes
theoretically directly contributes to the quality dimension.
Figure 30: Quality Terminology
Quality Dimension 1
Quality Dimension 2
Quality Dimension X
Dimensions
Quality Attribute A
Quality Attribute B
Quality Attribute C
Quality Attribute D
Quality Attribute E
Attributes
Masters Thesis Chris Taylor
QUT Page 93
This simply provides use the terminology that is used throughout the rest of
the discussion.
4.3 Research Question
The research question driving this sub-study is:
What is quality in terms of business process reference material?
Quality needs to be defined from a particular perspective which is made up of
the view of a particular class of observers and the purpose of the use. Figure
31 shows the relationship between view, purpose, perspective and quality.
The figure indicates that a perspective is made up of a particular purpose and
a particular view, and that from each perspective the perception of quality
may differ. The important fact that this outlines is that changing perspective
(i.e. by changing the view, or purpose, or both) with which an entity is viewed
will change the perceptions of quality about that entity.
Figure 31: Relationships between view, purpose, perspective and quality
The perspective examined in this chapter is from those who are using the
model for business process re-engineering. Therefore, the view is that of the
users (i.e. those directly applying the model). Although the user view is only
one type of perspective which is needed to define quality, it is believed that
this view is an important one, as the users of the model are effectively the
customers of the reference model producers. This view is of those who are
Perception of Quality
Perspective
determines
View (user)
Purpose (bpm)
Is part of Is part of
Masters Thesis Chris Taylor
QUT Page 94
involved with the application of the reference model, i.e. those responsible for
the “Use” phase of the reference model lifecycle, as shown in Figure 32.
Figure 32: RM Lifecycle phase in which the Quality research is based
The purpose of the reference models selected by the organisations involved
with this research is to aid in business process modelling, and these models
were used for this purpose.
In this research quality is defined as:
The degree of existence of user desired attributes.
This definition is consistent with other definitions of quality which have been
collected and summarised by Ivancevich et al. (1997 p.10).
4.4 Research Design
First, a comprehensive review of the relevant literature was conducted
drawing upon related fields namely reference models, conceptual model
quality and business process modelling. From this the quality framework was
derived. This framework provided the basis from which to draw questions
about specific attributes of quality. User perceptions and reactions to the
proposed quality attributes were then tested within two case studies where
business process reference models were extensively used.
Design Reference
Model
Reference
Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 95
The primary objective of the case studies was to identify those attributes that
were perceived by the users as important for reference model quality.
4.5 Quality Framework
Several existing popular quality frameworks are presented here as a basis for
the derivation of the quality framework used in this research.
4.5.1 Quality of Modelling Framework
Lindland, Sindre and Solvberg (1994) published a paper outlining a
refinement of quality aspects of a conceptual model. The work was heavily
influenced by the then popular data models. They helped to separate the
sometimes jumbled concepts of the definitions of quality and provided a
simple framework for the three types of quality. They defined four important
entities with relation to the quality of a model.
The first, the model itself, contains a number of statements. Second, the
audience interpretation is the set of interpretations that the audience has
made based on the model. The third entity is the domain, which consists of
all possible statements that are correct and relevant to a particular problem.
Lastly, the language is all the possible statements that would be possible
given a particular set of language rules (i.e. the syntax).
The quality of the model is derived from the relationship between these
entities. The qualities defined are syntactic, semantic and pragmatic quality.
Syntactic quality is the correspondence of the model with its language, in
simple terms “the more closely the model adheres to the language rules, the
higher the syntactic quality” (Lindland et al. 1994 p46).
Semantic quality is the correspondence of the model with its language. They
defined two types of semantic quality, firstly validity, which suggests that all
the statements in the model are correct and relevant to the problem.
Secondly, completeness means that the model contains all the statements
about the domain.
Masters Thesis Chris Taylor
QUT Page 96
The pragmatic quality is the correspondence of the model to the audience.
Pragmatics is concerned with alternative ways of expressing the same
semantics. The goal for pragmatic quality is comprehension, when all the
model statements have been interpreted correctly by their relevant audience.
The Quality of Modelling (QoM) was also extended by Krogstie et al. (1995),
to include Physical model, which is the degree of persistence and availability
of the statements, Language quality, how well the language suits the domain,
Social quality, which is how well the audience agrees about the model.
4.5.2 Guidelines of Modelling
Becker, Rosemann and Uthmann (Becker et al. 2000) put forward their
guidelines of business process modelling outlining 6 quality factors of
business processes models. In their work on extending the definition of
quality beyond data modelling and focusing on business process model their
goal was to extend the concept of model quality beyond the notion of
syntactic correctness. The factors were correctness (of semantics and
syntax), relevance, economic efficiency, clarity, comparability and systematic
design.
4.5.3 Semiotic Framework
The semiotic framework (Stamper 1992; Falkenberg et al. 1998) is not a
quality framework as such. It is a 6 layer model that describes
communication. It provides a good model for understanding the various
levels of communication and how they interact. It is discussed here because
it provides a broad scope of communication and may suggest qualities that
may not have been already covered by the two quality frameworks. The 6
layers are the physical, empirical, syntactic, semantic, pragmatic and social.
4.5.4 Deriving the Quality Framework
The framework and dimensions are derived to provide a theoretical basis and
to ensure that all previously mentioned aspects of quality have been included
in the case study questions. They served as categories in which the
Masters Thesis Chris Taylor
QUT Page 97
assimilation of the quality attributes defined by various authors could be
performed.
The overlap between the three frameworks and the derived framework is
presented in Table 15. The first column defines the new proposed terms, the
second, the Quality of Modelling (QoM) combines the qualities defined in the
Lindland et al. and the Krogstie et al. papers. The Guidelines of Modelling
(GoM) summaries the guildlines proposed by Becker et al. and finally the
semiotic framework summaries the aspects of communication.
Proposed Terms
QoM (1994 + 1995) GoM Semiotic
Framework
Syntactic Correctness Syntactic
Comparability
Systematic design Syntactic Quality
Language Quality
Semantic Correctness Semantics Semantic Quality
Relevance
Pragmatic Clarity Pragmatics
Physical Quality Physical
Empirics
Pragmatic Quality
Social Social Table 15: Overlap in the Modelling frameworks and proposed quality dimensions
The proposed quality dimensions have been made by interpreting the
frameworks and grouping the similar concepts. For example, Lindland et al.
(1994) and Becker et al. (2000) both identify that having semantic
correctness is an important quality aspect of a model and a business process
model respectively. This also maps onto the semiotic framework’s
“semantics”.
We define 3 dimensions of quality for reference models; Syntactic, Semantic
and Pragmatic as shown in Figure 33. Although this seems to be a direct
replication of the Lindland et al. framework, the definitions of the quality
dimensions is expanded from the original to include influences from the
Guidelines of Business Process Modelling (Becker et al. 2000), Semiotic
Masters Thesis Chris Taylor
QUT Page 98
framework and the revisions to Lindland et al.’s original work in (Krogstie et
al. 1995) and other papers. The new definitions are discussed next.
Figure 33: Quality Framework for this paper
Syntactic Quality
The languages used in business process reference models are often not
languages with strict, formally defined rules. Therefore there is no formal test
for compliance or non-compliance with a set of language rules is possible.
Rather, the syntactic correctness relates to the properties that describe the
structure and organisation of the model itself as in Misic and Zhao (2000).
Often reference models do not have a defined meta-model or accompanying
modelling conventions, hence judgements must be made about the implied
meta-model and conventions. By defining syntactic quality as such, we are in
effect broadening the definition of language to include not only the implied
meta-model but also the implied layout, overall design and underlying
concepts that have been applied to depict the content of a model.
Using this definition of language, we see that syntactic quality is not only the
strict applications of the grammar and symbols, but also how these are used
in terms of consistency, especially with respect to aspects such as layout,
naming conventions etc. It also includes the mapping of the language and
constructs used to depict the real world entities as can be explained in
glossaries or ontologies for example. This definition of syntactic quality
covers Becker et al.’s (2000) “(Syntactic) Correctness”, “Comparability” and
parts of the “Systematic design”. Our definition maps to the semiotic
framework’s “syntactic” layer.
Language
Audience
Domain Model Syntactic Semantic
Pragmatic
Masters Thesis Chris Taylor
QUT Page 99
Semantic Quality
Semantic quality reflects how well the model captures what it is supposed to
capture as defined by the model scope statement or implied scope
statement. An ideal fit in this case is that the model correctly captures
everything of relevance and nothing of irrelevance. This is essentially
Lindland’s et al. (1994) original definition of “completeness”. This quality
dimension is similar to Becker et al.’s (2000) “(Semantic) Correctness”, and
“Relevance”. Our definition also subsumes Zamperoni and Lohr-Richter’s
“consistency” where no model statement contradicts another model
statement (Zamperoni and Lohr-Richter 1993). This is the situation where,
although the language has been applied correctly, two statements in the
model are contradictory. If two statements are contradictory then at least one
of the statements must be incorrect, hence we have classified this as a
semantic error. Semantic consistency is also mentioned as important in
semantic quality in Misic & Zhao (2000). Our semantic quality maps to the
semantic layer of the semiotic framework.
Pragmatic Quality
Again expanding from Lindland et al.’s (1994) definition, pragmatic quality is
defined here as how well the model is understood, which includes the quality
of the model itself and the quality of the support material e.g. training,
explanations etc. “Perfect” pragmatic quality would mean that every part of
the model is correctly interpreted by the relevant audience and that this
interpretation provides value to the audience. This quality dimension also
corresponds to Becker et al.’s “Clarity”. As a means to this understanding,
the distribution of the model is also important; hence pragmatic quality also
includes attributes of the semiotic framework’s physical layer. Our reasoning
being that an audience cannot interpret that which it cannot access. It also
includes attributes from social quality from Krogstie, Lindland and Sindre
(Krogstie et al. 1995) and the semiotic framework. Social quality was
included in the pragmatic quality because, assuming the model has only on
correct interpretation, any inconsistencies in individuals interpretations, which
is a lack of Lindland et al.’s social quality, must be due to at least one
incorrect interpretation. This incorrect interpretation is therefore an
Masters Thesis Chris Taylor
QUT Page 100
expression of poor pragmatic quality of the model; hence social quality is
dependent on pragmatic quality and is included within it for this discussion.
Feasibility
One aspect that was left out of our framework that has been repeatedly
mentioned in literature is the concept of “feasibility” (Lindland et al. 1994) or
“economic efficiency” (Becker et al., 2000). This type of quality weighs the
cost/benefit of further improvements, and asks the question “does the cost of
incrementally improving the model out-weigh the benefits gained from such
an improvement?”. At the point the cost is higher, the literature suggests that
the optimum economy efficiency has been achieved. This concept is solely
focused on the design of a model. From the user requirements point of view,
the reference model producer’s effort is irrelevant and probably unknown and
as such we excluded it from the case study framework.
Framework Summary
The quality framework is the theoretical basis we have developed to ensure
that we ask the right types of questions about previously mentioned quality
dimensions and attributes and to provide a high level view for the data
analysis. The major outcome of this research is the list of specific quality
attributes identified from the case studies.
4.6 Case Study Descriptions
The reported study was conducted in two organisations: Queensland Rail
(QR) and Telstra – Australia. Contained here is the context of the usage of
the reference models within these organisations. For a brief description of
these organisations see the Appendix E.
These organisations were involved in the research because they undertook
process modelling activities using business process reference models, and
were accessible to the research team.
Today, modelling is extensively used at QR within different projects and for
multiple purposes. The Supply Chain Operations Reference Model (SCOR)
(Anonymous 2002) was extensively applied within one of QR’s process
Masters Thesis Chris Taylor
QUT Page 101
improvement research projects; the Rail supply chain optimisation project.
The Rail supply chain optimisation project was proposed as a strategic
initiative to develop a methodology that QR managers can use when deciding
how to manage and improve the performance of their inbound supply chains.
Supply chains require all parties to move in unison. Process modelling came
into being within this project to address this need – to have a standardised
way to exchange information, within a wide variety of stakeholders of a
supply chain.
Telstra is heavily involved with the development of eTOM and is working on
using the framework to develop standards both within the company and in its
interactions with others. Telstra’s adapted version of eTOM, is named
TeTOM. Telstra uses TeTOM for several purposes, including the modelling of
the internal processes and as a method of identifying duplication of
processes, or resources to support these processes. Telstra is also looking at
using the reference model for process improvement projects.
4.7 Data Collection
A comprehensive case study protocol was derived (see Appendix A),
documenting all procedures relating to the data collection and analysis
phases of the study.
Data was collected from 5 respondents (2 from Telstra and 3 from QR) from
the case sites in the form of semi-structured interviews. All interviewees had
extensive experience in using business process reference models. The
questions were drawn from the literature using the framework discussed in
the first part of this chapter. General questions about the identified quality
dimensions were followed by a series of questions testing the individual
attributes of quality. After the interviewees had responded to each question,
they were prompted to comment on the importance of these attributes based
on their experience. Other sources (e.g. project documentation) were not
studied.
Masters Thesis Chris Taylor
QUT Page 102
4.8 Data Analysis
Analysing case study evidence has been a noted challenge by most case
study experts (Yin, 1994, p.102). Only a few case data analysis techniques
and supplementary tools and techniques for data analysis have been
discussed in literature (Miles and Huberman 1984; Yin 1994). We first briefly
discuss the method we applied to analyse the case data. The next section
presents the findings with evidence.
“Codes are tags or labels for assigning units of meaning to the descriptive or
inferential information compiled during a study” (Miles and Huberman 1984 p.
55). A comprehensive data analysis tool; Nvivo 2.0 was used for the data
analysis of this study. Data were coded within the tool to elicit meaning and
present summaries of the phenomena investigated. A tree like node structure
was initially created to capture the quality factors of the a priori model;
namely Syntactic, Semantic and Pragmatic quality. Whenever a quality
attribute was mentioned or hinted at, it was coded with the relevant
node(s).The data coded under each code was re-analysed to make sure that
they did belong to the coded quality construct. The next phase re-analysed
the data coded under each node and recoded them at a further level,
differentiating between instances that discussed the weak or strong existence
of the quality factor and potential means of achieving the quality factor. In the
final phase of data coding, In-vivo (coding bottom-up, by using the key words
from the data) coding was conducted to derive a list of sub constructs that
described each quality attribute.
The codings were conducted by one coder (Sedera), and then verified and
modified after discussion by the interviewer (Taylor).
4.9 Findings
The main purpose of this research was to simply identify the quality
attributes, not to provide a relative ranking. Some indication to the relative
importance of the quality attributes can be obtained by comparing the
number of citations for each individual quality attribute across the different
case studies. ‘Numbers’ usually get ignored in qualitative research; however
Masters Thesis Chris Taylor
QUT Page 103
a lot of counting actually does take place in qualitative studies when
judgements are made. For example, we “identify themes or patterns that
happened a number of times and that consistently happens a specific way”
(Miles and Huberman 1984 p.215). Thus, Table 16 presents not only the
different attributes for each quality dimension, but also the number of times
that they were mentioned as important across the case studies. This can aid
the interpretation of this study and provide a preliminary impression of the
relative importance.
Also of importance is the context with which the attributes were discussed. In
this section we interpret the Table 16 and also elaborate on the context,
presenting an accurate description of the case study findings. The results of
the case studies have been combined in the table, because responses were
generally uniform and we did not attempt any cross case analysis.
Syntactic Quality Semantic Quality Pragmatic Quality
Attributes cit. Attributes cit. Attributes cit.
Language must be clearly defined 14
Clearly define the scope of the model and its
limitations 10 Application Documentation 14
Language must be used consistently 14 Prior validate the models 5
Model users must be educated about the
reference model 10
The language must be simple to understand
7 Obtain feedback from stakeholders 5 Provide tool support 6
Have a glossary to define the key terms 7 Models must be
semantically consistent 4 User should be able to
maintain contacts with the vendors for clarifications
6
Provide Vendor training; explaining
the terminology 5 Avoid unnecessary
repetitions 4 The model must be easily accessible 3
All levels of the reference model must be defined
adequately
3 Try to capture the entire domain as a whole 4
Provide supplementary material to support user
understanding of the content 3
Create a meta model 3
Have realistic and feasible content (I.e. recommendations)
1 Model should be tailor-able to specific needs 2
Avoid the use of unfamiliar terms 2 Only have relevant
statements 1 Conduct vendor training on
the conceptual aspects of the model
1
Models must be laid out properly 2
Table 16: Citations of Quality attributes from case study interviews
Repeatedly mentioned was the need for a clearly defined and consistently
applied language. This point was raised often, and sometimes in response to
Masters Thesis Chris Taylor
QUT Page 104
questions not dealing with the language. Interviewees consistently expressed
strong opinions on the importance of consistent clear language, e.g.
“extremely important”. In the words of one respondent the lack of consistency
“discredits the … quality” of the model. Respondents mentioned the
importance of the consistency of language use fits neatly with the emphasis
placed on consistency in the literature.
All the respondents indicated that the ability to ask questions and receive
answers was an essential part of using a reference model. These question
and answer sessions, delivered in a training situation or a simple email
conversation, were variously described as “very important” and “extremely
important” and saved users “a tremendous amount of time”.
A deviation from literature is found on the issue of relevance. Most existing
model quality frameworks indicate that a model is of low quality if it contains
irrelevant material. The results from the case studies indicate that having
material that is not relevant to the audience may not negatively impact on the
model’s perceived quality (i.e. only having relevant statements was
mentioned once, while capturing the whole domain was mentioned 4 times).
Most respondents indicated that at least part of the models they were using
were not applicable to their situation. However, this was not identified as a
quality problem. This may be an important distinguishing feature between
quality of specific models and reference models. It is a conclusion of this
research that a reference model has different requirements in terms of
completeness.
Again the issue of clarity in the definition and use of the model language,
especially process naming, is essential in this case to identify irrelevant parts
of the model, evidenced by comments such as “having the definitions for the
bits that weren’t relevant to us was essential”.
A model that is “overly-complete”, that is, it covers a domain that is beyond
the scope of a particular project or particular audience would require only
red-lining, i.e. deleting the non-necessary pieces (as compared to the other
methods of derivation, configuration and embellishment), while a reference
Masters Thesis Chris Taylor
QUT Page 105
model that was not sufficiently complete, would require the addition of
sections of the model. Deleting parts of the model is much easier than
adding. Also, red-lining maintains the original syntactic and semantic
consistency of the model, because no mistakes are introduced to the model
in contrast with when the model users add new parts.
Another potential reason a reference model is of higher quality if it is overly
complete is the scalability of the model, meaning at later times more of the
model can be utilised, although this was not specifically discussed by the
respondents.
Linked to this issue of completeness was the result that users desired a
reference model to capture an entire domain, in one case an entire
telecommunications service provider as a commercial entity and in the other
a complete supply chain. A reason for this could be that a reference model
provides standardised terminology in what can be large and complex
domains, as is the case in these case studies. Sacrificing conciseness for
completeness may have a positive effect on user perceptions of the quality of
the model, by ensuring that the complete large and complex domain has
been mapped in its entirety, even if it is not used in such a widely scoped
modelling project.
One interesting concept that was canvassed in the interviews was the use of
pragmatic variation (or individualisation) in the models. That is, the idea that
the same underlying model would be presented to different stakeholder
groups, but with the terminology changed to suit the intended audience. This
strategy was raised during the case studies and is an important new concept
that could increase the pragmatic quality of a reference model.
A terminology matching dictionary or functionality could be developed by
reference model providers for various expected audiences, especially to use
terminology that is common in other reference models or for different
perspectives. This could be as simple as a table listing the terminology
matches between the model, other common models or common terms for
various stakeholder groups, e.g. technicians, managers etc. An interesting
Masters Thesis Chris Taylor
QUT Page 106
example from the QR case was the initial confusion about the Americanised
terms as opposed to Australian terms used in the model. Feedback from the
case studies was that the use of terminology customisation could defeat the
standardising effect of a reference model. This example reflects the comment
of Gulla and Brasethvik’s (2000) that variation of models complicates the
formation of a common understanding of the domain. Depending on how
broad the scope of the standardisation of terms is designed to be, the
overuse of the terminology matching dictionary could still be of great use. For
example, the use of terminology variation might not be appropriate if the
major aim of the model is to align a whole industry, but may well be useful to
match the terms to that of a single organisation that is using the model.
The coding in Table 16 and the context in which the models were discussed
confirms the importance of economic efficiency in the quality attributes or
dimensions of reference material. Quality aspects such as Application
Documentation, easily understood, easily accessible, have a simple
language etc are all examples of quality attributes that can aid the economic
efficiency. The desire for guidance on how to use the models (e.g.
implementation guides, case studies) and aspects that would reduce the time
and effort to understand and communicate the reference model (e.g. simple
to understand language) support this “economic efficiency of use” concept.
Our findings indicate that an important quality attribute or even a quality
dimension from the viewpoint of the user of a reference model is the
economic efficiency of the use of the model, a separate concept to the
previously mentioned economic efficiency of the design of the model. This is
a separate economic efficiency to that mentioned in the literature, which
discusses the economic efficiency of the design process; the two concepts
are not necessarily related.
The relationship of these concepts is depicted in figure 34.
Masters Thesis Chris Taylor
QUT Page 107
Figure 34: RM Lifecycle related to the types of economic efficiency
4.10 Limitations and Conclusions
A limitation of this research is external validity, with only 2 case studies. This
chapter has presented the empirical findings of two case studies aimed at
determining the quality attributes of business process reference material.
This research is not, and is not intended to be a definitive study on the quality
of business process model. It is only an exploration of the user perceived
attributes of quality. It is intended to provide a basis for further work in the
area. Another limitation is the use of the frequency of citations as an
indication of the perceived importance of the various attributes. Again this
should be taken as indicative only and requires further work to test the
relative importance and completeness of the attributes. The use of
perceptions of quality only and not other more objective means such as direct
comparison is another limitation. However, direct comparison should be done
on proven quality criteria, derived from exploratory research such as this
paper. Also as mentioned in Krogstie et al. (1995) it is difficult to measure
quality directly, particularly semantic and pragmatic quality and as a
substitute for direct measures perceptions are a valid measure.
In conclusion the consistent use of terminology, a simple or simply
understandable modelling language, and a complete, accurate and ideally
tested, content were all results that had been mentioned previously in the
literature, although not extensively empirically explored or proven. Findings
Design Reference
Model
Reference
Model
Use Reference
Model
Efficiency of design
Efficiency of use
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 108
that were not mentioned in the existing literature or contradicted it included
the issue that reference models should be complete at the expense of being
relevant to a particular audience, and the need for a feedback mechanism
either through training or some other means. An interesting new idea
expressed throughout the case studies was the idea of customisation of the
terminology used in reference models. The original framework was found to
be lacking in that it did not contain aspects of economic efficiency,
particularly of the cost of use of the reference model.
Based on these results, further exploration of the quality of reference models
should be conducted particularly from differing points of view (e.g. from the
model producer’s view). In this further research we would expect that the
economy of design and, particularly if the reference model is living, the
economy of maintenance to be important quality factors.
Figure 35 summarises the conclusions that differ from or add to the existing
literature on the quality of models using the particular perspective in this
research. The major reasons for the differences found in this research to the
current literature probably stem from the slightly different unit of analysis (i.e.
reference models versus models in general) and the perspective taken on
quality (i.e. user view and business process modelling purpose versus
general quality drawn largely from modelling for the purpose of designing
computer systems).
Figure 35: Conclusions that add to or differ from current literature
Syntactic Quality • Glossary • Simple
Semantic Quality • Scope and
limitations • Validation of
models • “Over-
completeness”
Pragmatic Quality • Application
documentation • Vendor training • Feedback
mechanisms
Economic Efficiency of Use
Masters Thesis Chris Taylor
QUT Page 109
The choice of case study organisations may also have been a limitation to
the external validity of the findings. The two organisations were engaged in
process modelling, however in both cases these modelling projects were not
very large in scope or size and did not use a formal modelling technique or
dedicated tool. As such the lack of language formality (as defined in
Chapter 2) in the models may not have presented any problems as it may
have done if the modelling efforts were conducted in a more systematic and
syntactically formal manner. The models were used for a semantic basis
only, not as true templates, so the aspects of quality may be different
depending on the type of derivation. Because the models were not used for
an IT implementation, the demands for a formal language were also reduced.
Further work in other case studies or a survey encompassing a larger sample
would be required to validate the findings of this study. We believe that this
paper has laid the exploratory foundation for further empirical research into
the area.
4.11 Chapter Summary
This chapter has explored the quality of business process reference models
from the perspective of those using the reference models for process
modelling. It is designed to offer reference model producers guidance on
what factors are important to the perceived quality of their reference material
as perceived by the model users. As mentioned in Chapter 3, the case
studies and analysis were concluded in parallel. The study findings benefit
future research in the domain of reference models and more specifically on
reference model quality. The study findings is of value to the practitioner who
applies business process reference models, as the findings can be used in
the selection and evaluation processes of reference models.
Masters Thesis Chris Taylor
QUT Page 110
Chapter 5: Reference Model Design Procedure
5.1 Introduction
This chapter outlines the proposed procedure for the design of a reference
model.
Figure 36: Chapter 5 in relation to the Reference Model Lifecycle
The term procedure is used here to indicate a project (as differentiated from
a continuous or operational process.
There has been very little academic literature outlining a procedure to design
a reference model. In fact, there were no articles specifically describing the
procedure of the design of a reference model, although several mention the
time and effort required to design a reference model.
An example from Germany gives an indication as the time and resource
commitments necessary. The construction began in April 1999, consisted of
seven project subgroups, several steering committees, and more than 100
university employees with a total of 30 consultants and instructors from
Mummert & Partner, Debis, and SAP SI. One of the project goals was to
develop a reference model for financial processes at German universities
and 2 years later went live with only parts of the FI and Funds Management
modules (Anonymous 2001).
Design Reference
Model
Reference
Model
Use Reference
Model
Classification Characteristics
Quality Attributes
Masters Thesis Chris Taylor
QUT Page 111
Inferences can be drawn as to the effort and personnel required to produce a
reference models by the number of authors. For example reference models
such as SCOR, eTOM and ITIL have up to several hundred listed authors
and many more contributors.
As can be seen by these numbers the design of reference models can be a
long, and therefore presumably expensive, procedure involving many man-
weeks.
The next section identifies the design philosophies for reference models. In
reality the creation of a specific reference model may be based on many of
these philosophies, however an explicit description of each may clarify the
options available to reference model designers.
5.2 Reference Model Design Philosophies
Having found no detailed theory on how to design a reference model, several
design philosophies, or methods are proposed.
We also propose that the design philosophies chosen for a particular
reference model design should be influenced by the purpose and intended
audience of the model and proposed characteristics of the model, particularly
the Readiness for Use (e.g. single depiction, contains variant) and State (i.e.
best or common practice) characteristic. The design philosophies are
presented here as formalised and discreet approaches. However, in reality,
the actual design of a reference model would consist of many trade offs and
fluid combinations of these approaches. Indeed, the decisions described in
this section may not be made consciously or explicitly, but rather are part of
the “black-box” of reference model design.
Many of these design philosophies rely on the accurate identification of the
appropriate existing enterprise specific models, reference models or other
information (collectively termed the information entities for this discussion,
see Figure 37 for examples), and the accurate comparison of these
information entities to determine which parts of the entities correspond to
each other and whether each part is relevant to the reference model being
designed. These three steps, identifying relevant information entities,
Masters Thesis Chris Taylor
QUT Page 112
identifying the relevant parts (or sub-entities) of these entities, and comparing
these sub-entities, could be particularly difficult and may not be based on
scientific or explicit manner. This design procedure is essentially creative
and, as such, often unstructured, relying heavily on the modellers’
interpretations of the information entities, other subjective decisions and the
pressures of time and resources available during design.
Figure 37: Examples of possible information entities
Each of the philosophies is identified below and their advantages and
disadvantages are briefly described.
5.2.1 Blue Sky Design
This design philosophy is extremely creative and is based on the original
concept of BPR. It intends to design a reference model without relying on
other sources, such as the existing models in the domain. Reasons this
philosophy may be used include the absence of any other information, or a
desire to be freed of the constraints on creativity sometimes associated with
going through the design process with existing ideas and/or constraints. This
process could be challenging and may ignore existing good applicable ideas
or unavoidable environmental constraints; essentially it may be a case of re-
inventing the wheel. Alternatively, the resultant reference model may contain
unconventional, visionary, and innovative ideas.
Reference model to be designed
Case Studies
Surveys
Tacit Knowledge
Focus Groups
Other Reference
Models
Academic / Trade Papers
Existing Products
Standards
Masters Thesis Chris Taylor
QUT Page 113
A model of this type may serve as the ideal long-term goal within an industry
or organisation.
5.2.2 Design by Choice
There are two types of design by choice. Firstly, a model may be declared as
a reference model because it is seen as the best in the domain. The second
type of design by choice is the arbitrary selection, when the choice is
influenced by external forces that effectively dictate which model is the
reference model. An example of this, linked to the implementation of
Enterprise Systems, would be the use of a model of head office processes to
design and implement the processes at any subsidiaries or local offices in an
international organisation. This could enforce or improve standardisation
efforts which are based on a known working implementation.
Both of these types of Design by Choice transfer the status of “reference
model” onto an enterprise specific model. This method is commonly used
when organisations are identified as having “best practice” in a particular
area, where their model is taken as the reference model for that area, which
could then be used for applications such as benchmarking.
5.2.3 Baseline Design
To produce a reference model using the Baseline design approach, several
steps are required. Firstly, information from various sources detailing the
target domain would need to be identified. These information entities could
include other models, “expert” opinions, academic and trade literature,
standards or any other definitions or representations.
Once the entities have been identified they have to be related to match like
concepts within the information entities. Once this identification and
comparison of these information entities has been completed, the existence
of these sub-entities can be related. In the Baseline Design approach like
sub-entities that did not appear in sufficient numbers across the information
entities could be excluded.
Masters Thesis Chris Taylor
QUT Page 114
Using this approach the identification of suitable information entities is crucial
to the outcome, as is the selection of the reference model scope. The
resultant reference model reflects the maximum commonality of the domain
or in mathematical terms the highest common denominator.
Figure 38: Baseline Design
In the example shown in Figure 38 the sub-entity must occur 80% or more of
the information entities for it to be part of the reference model.
5.2.4 Common Practice Design
Similar to the Baseline Design, this approach starts with the identification and
comparison of the information entities and considers the number of times
sub-entities appear in the information entities. The highest occurring sub-
entity is then chosen for the reference model.
Reference Model Relevant Parts of Information Entities
Masters Thesis Chris Taylor
QUT Page 115
Figure 39: Common Practice Design
Advantages of this philosophy include the fact that the reference model
reflects the most common implemented solution for a particular domain. This
could then be used for standardisation activities. Many parties will be able to
relate to the reference model, which could make it more generalisable. A
reference model designed this way does not contain any radically new ideas.
5.2.5 Best Practice by Composition
Again starting with the identification and comparison of the relevant
information entities, this philosophy is an attempt to provide a “best-of-breed”
approach to the reference model. When information entities are in
disagreement a method (for example case study, statistical analysis of
implemented occurrences, survey, experience) could be used to determine
the “best” solution. This may be a matter of choosing the best solution or
combining existing solutions to produce the best practice. This method of
composing sections of existing information entities to form a reference model
may be extremely difficult and subjective as the information entities may not
have clearly defined points which could aid integration, and often may not be
depicted in the same modelling language. This would involve a certain
amount of interpretation on the part of the model designers.
Reference Model Relevant Parts of Information Entities
Masters Thesis Chris Taylor
QUT Page 116
Figure 40: Best Practice by Composition
Figure 40 depicts how certain sub-entities can be combined in a reference
model to capture best practice by composition.
This method could involve merging information entities to produce a new
idea, which is depicted graphically above in the second reference model
object. Advantages of this method is that new and innovative ideas can be
built into the model, although disadvantages may include the effort involved
in design, and the fact that the generalisability of the reference model may be
low because of the cost of implementing all the best practice ideas, for
example several of the best practice information entities may be built with
different IT systems, making the integration of these point solutions a
requirement for the implementation of the reference model in its entirety.
5.2.6 Design by Abstraction
This approach requires a set of relevant information entities to be identified.
After the comparison of these entities, the Design by Abstraction approach
would create the reference model at such a high conceptual level that all the
variations are accounted for within the higher constructs.
Reference Model Relevant Parts of Information Entities
Masters Thesis Chris Taylor
QUT Page 117
Figure 41 shows a graphical depiction of this design philosophy, the blank
square is a depiction of “a symbol” which is the abstraction from the circles
and hexagons in line 4.
Figure 41: Design by Abstraction
Figure 41 depicts how the reference model can be abstracted from the
Information entities, by capturing only what is common in all the information
entities.
This approach produces a reference model that is very general, and could be
applied to many circumstances, but one with lacks the level of detail of the
original entities. It requires the model designer to decide the point at which
the advantages of having a highly generalisable model are outweighed by the
disadvantages of a model lacking lower level detail, as depicted in Figure 42.
Reference ModelRelevant Parts of Information Entities
Masters Thesis Chris Taylor
QUT Page 118
Figure 42: Relationship between Benefit vs. level of abstraction
The exact curves on the graph are determined from the intended use and
audience of the model.
An apparently obvious solution of combining both the high level models
created by this method and lower level models created using any of the other
models does not adequately solve the problem. This is because the lower
level (higher detail) models require more specific context, decreasing the
benefits from generalisability. The higher level model could be used widely,
but the lower level models could only be used in a specific domain. What is
advantageous is a model with a single high level picture, but with several
solutions at the lower level that can be selected to fit the specific domain, that
is, an Explicit Alternatives Design.
5.2.7 Explicit Alternatives Design
This philosophy uses the same identification and comparison of information
entities as described above. In this case however, differences between the
information entities are captured in the model in an explicit way, allowing the
reference model user to select the appropriate alternative when instantising
the reference model.
Level of Abstraction
Benefit of Generalisability
Benefit of Level of Detail
Ben
efit
Masters Thesis Chris Taylor
QUT Page 119
Figure 43: Explicit Alternatives Design
The explicit alternatives design philosophy provides an interesting benefit of
combining the benefits of being widely acceptable, while still containing the
detail to make it useful at a lower level of abstraction. Disadvantages include
the cost to produce such a model and the fact that the model is not “ready-to-
run” i.e. it requires configuration before it makes sense at an enterprise level.
This second disadvantage could be over come if the model had default
configurations. At present this only relatively simple examples of this type of
feature has been found in existing reference models (e.g. SCOR), using a
functional explicit alternatives, but not any process examples of explicit
alternatives.
5.2.8 Design Philosophy Summary
Most of the philosophies presented here could also involve some aspect of
weighting of different information entities, based on information such as
timeliness, relevance, perception of excellence or trends.
The weighting could help reference model developers quantitatively skew the
design procedure to ensure the reference model focuses on the important
entities. An example situation where this could be advantageous could be a
software developer developing a reference model for the development of a
Reference Model
or
or or
Relevant Parts of Information Entities
Masters Thesis Chris Taylor
QUT Page 120
new off-the-shelf product, but wanting to make sure that the product
adequately supported the processes of its more valuable customers.
5.3 Reference Model Design Procedural Model
With the absence of any guidance, the reference model design procedural
model was designed as depicted in Figure 44. The model was based on the
general procedure for process modelling gaining validation for each model
while increasing the level of detail. In any modelling project, including
producing a reference model, the procedure needs to be flexible. This is
particularly true with respect to the modelling conventions.
In the following pages each of the steps of the proposed procedure are
described.
Masters Thesis Chris Taylor
QUT Page 121
Figure 44: Proposed BP Reference Model Design Procedural Model
DesignHigh-Level
ValidateHigh-Level
TestHigh-Level
DesignLow-Level
ValidateLow-Level
SelectLow-Level
TestLow-Level
No changesmade
Low Level OK,RM finished
DefineReference
Model
Changesmade
Low Levelchanged
Low Level OK,RM notfinished
Reference Model to be
designed
Masters Thesis Chris Taylor
QUT Page 122
5.3.1 Define the Reference Model
Outline Purpose of Model
This is closely linked to the purpose of the model. Not only is defining the
purpose of the model extremely important for the users of the model, it is also
important as a guiding point when creating the model. Closely linked to
purpose is the clear definition of the scope and limitations of the reference
model. This is one of the quality factors that is rated very highly as discussed
in Chapter 4.
Select Model Characteristics
The selection of model characteristics is important for several reasons as
explained in 2.3.
Firstly, it allows the developer to position the model with respect to any
existing works, therefore possibly allowing the reuse of other model contents
or structure. It also provides a clear view of the end product of the reference
model to allow consistent model design.
Many factors are important in choosing the characteristics. The three most
important aspects are the desired end use of the model (the purpose), the
proposed audience of the model and a group of considerations which is
termed the ‘business factors’.
These business factors include how or if the model production costs are
recouped, how the model is distributed and what support in terms of training,
feedback capabilities etc. are provided; essentially the business plan behind
the reference model. These factors are not discussed in this thesis.
Develop Modelling Conventions
Developing modelling conventions reduces the numerous different possible
applications of the modelling technique. Modelling techniques or languages
usually only contain a part of the rules required for a consistency in the
layout, terminology, naming and content. These types of consistency have
been shown to be amongst the most important quality factors for end-users
of reference model as discussed in Chapter 4.
Masters Thesis Chris Taylor
QUT Page 123
During this stage, the terms used in the model should be set up and
constantly updated during the rest of the reference model design to reflect
any changes made. This provides a glossary of the terms which was
identified as a key quality attribute for reference models in Chapter 4.
5.3.2 Design
The design step is dependant of the design philosophy or philosophies that
have been chosen by the developers, which has been discussed in 5.2.
Although these steps are described as discreet and calculated activities, the
actual design of a reference model may not be a linear and well structured
process. The steps as described here are only meant as a high level
simplified view of the activities that could be required.
Table 17 summaries which steps would be needed dependant on the design
philosophy used.
Steps
Identify
information entities
Relate information
entities Rate entities
Blue Sky
Abstraction + +
Choice + +/- +/-
Baseline + + +
Common practice + + +
Best practice (composition) + + +
Philo
soph
y
Explicit Alternatives + +
Table 17: Relationship between Design Philosophy and required Design Steps
+/- This means that the step may or may not be required
Masters Thesis Chris Taylor
QUT Page 124
Identify Information Entities
This step involves searching for information related to the reference model.
The information collected in this stage could include tangible, e.g. other
reference models, publications from standards organisations, academic and
industry literature and intangible resources, such as experienced individuals
or organisations. Relevant information entities may include material from the
same industry as the reference model, but may also include input from
similar areas, such as using the best practice for hotel check-in for hospital
admissions.
This stage may also include searching not just for the content, but for ways of
depicting the content, e.g. languages, modelling techniques, terminology or
popular graphical depictions or ways of structuring the information about the
domain.
Once an information entity has been identified, the entire entity, or only
selected parts of the entity (the sub-entities) can be used as input for the next
step of relating the entities (or sub-entities).
Relate Information Entities
This step involves identifying which parts of the information entities (the sub-
entities) are within the scope of the reference model and then comparing the
models themselves by grouping the like sub-entities. Due to the complexity of
most domains, and of the infinite methods of capturing information, the step
of relating information entities can be quite difficult.
For example, parts of a reference model describing the processes for a
manufacturing enterprise may be compared to parts of a model for a services
firm. Relating the information sub-entities may involve realising that the
processes for “HR” in the manufacturing model cover a similar domain to the
“Personnel Management” part of the services model.
Masters Thesis Chris Taylor
QUT Page 125
Figure 45: Depiction of the design step of Relating Information Entities
Figure 45 graphically depicts how different sub-entities can be related to
each other. Once they have been related to each other the design philosophy
determines how to combine the sub-entities into the final reference model.
Rate Information Entities
Entities could be rated through benchmarking studies, surveys, or using
qualitative measures or subjective judgements. The type and depth of the
required comparison depends on the requirements of the situation and the
resources available to the task.
Depending on the design philosophy/ies chosen the results of the rating of
the sub-entities (or entities as a whole if the Choice philosophy is employed)
determines which parts or the existing materials are re-used or modified to
design the parts of the reference model.
Build
This step involves integrating all the required parts of the information entities,
which could include thoughts, other models, case studies or standards etc. If
the information entities are in different languages, which is highly probable,
then transformation into a single modelling language is required.
Information Entity A
Relationships between Information
sub-entities
Information Entity B
Masters Thesis Chris Taylor
QUT Page 126
5.3.3 Validate
Validation is an important step for both the accuracy of the model content
and the acceptance by a target audience. It is a way of reconciling the
opinions expressed in the model by the developers with outside experts.
Some models do this by a public release and invitation for comment e.g.
eTOM. It is also an opportunity to test the pragmatic quality of the model, by
introducing it to those who have not been involved in the design of the model.
The ability to provide feedback to the developers and an opportunity to be
educated about and question the model was also identified as important. The
validation step allows all these to happen.
Validation can be achieved by using the reference model in real projects and
assessing its strengths and weakness as perceived by the reference model
users and using these as a basis for improvement in the model.
At the end of the validation the model should be verified, i.e. ensuring that the
model conforms to the language rules and layout conventions etc.
5.3.4 Test
Prior testing of the model was an important quality attribute identified in
chapter 4. By testing we mean using the model for its intended purpose and
identifying the areas that need revision or that need to be added. Once these
areas have been identified they can be changed by the model developers,
ideally followed by a re-validation of the changes. The number of iterations of
this cycle depends on the “economic feasibility” as discussed in Chapter 4.
Outputs from the test phase could be the comments by those who were using
the model as well as the material e.g. enterprise specific Models that the
team produced. This data should then be examined for possible changes to
the reference model.
One of the quality factors identified in Chapter 4 is the need for the model to
be tested in a real life context, a platform that the testing phases provide.
Masters Thesis Chris Taylor
QUT Page 127
5.3.5 Select Low-Level
The step of selecting a lower-level process to model is reasonably arbitrary.
Business factors such as which process would be first implemented in an
organisation or for which process there was a market demand could
influence the decision. As discussed in Chapter 4, a reference model should
be complete, meaning that all the processes shown in the high level model
should eventually be modelled to a least the same level for consistency. The
level to which they are modelled depends on the intended audience and
application of the reference model.
5.4 Chapter Summary
This chapter outlines several design philosophies for business process
reference models, namely “Blue Sky Design”, “Design by Abstraction”,
“Design by Choice”, “Baseline Design”, “Common Practice Design”, “Best
Practice by Composition” and “Explicit Alternatives”. Advantages and
disadvantages of each of these philosophies are briefly outlined. These
philosophies can be combined depending on the requirement of the
reference model under design.
The second major topic of the chapter is a proposed procedure for the
development of business process reference models, showing the effect of
each of the design philosophies on the design procedure. This proposed
design procedure is applied in the next chapter to design a partial reference
model for IT service provision.
Masters Thesis Chris Taylor
QUT Page 128
Chapter 6: ITSP Reference Model Design
6.1 Introduction
This chapter describes the actual design of the reference model for IT service
provision. The testing of both the high-level and the lower level are presented
in the next Chapter.
As described in 5.1 the design of a reference model is quite time, labour and
knowledge intensive and due to time and resource restrictions of the
research team and invited participants only one cycle of the design
procedure was completed.
6.2 Define the Reference Model
6.2.1 Model Purpose
A reference model may be applied for difference purposes; however it is
necessary when creating any model to have a specific purpose in mind. Most
business process models have been mentioned in literature as a tool for
business process re-engineering or business process improvement (BPI) as
discussed in detail in 2.6 e.g. (Rosemann et al. 2003 p27). This reference
model is therefore aimed at supporting this BPI purpose. Specifically, it
should be useful as a process identification tool, a TO-BE template and to a
lesser degree an AS-IS template.
Masters Thesis Chris Taylor
QUT Page 129
Figure 46: Purpose of the developed ITSP Reference Model
6.2.2 Model Characteristics
This section outlines the characteristics that were chosen for the model,
along with a short explanation and justification.
View
As explained in the opening chapter, the process view was selected for the
reference model. The process view was selected because business process
management has been successfully applied in other areas and types of
organisations, for example manufacturing or accounting, but is still relatively
immature in the IT service provision domain (Williams 1997).
Language
A graphical standard language has been chosen as the basis for this
reference model, namely the extended Event-Driven Process Chains (eEPC).
Most models within IT service provision rely heavily on natural language,
indeed early process models have employed narrative text and/or structured
Instantise for Process Identification
High-Level Template
Process Targeting
Instantise for AS-IS Modelling
AS-IS Template
Analysis Process Benchmark
Instantise for TO-BE Modelling
TO-BE Template
Process Implementation
Process Execution
Monitoring/Control
Masters Thesis Chris Taylor
QUT Page 130
textual descriptions (Curtis et al. 1992). However, one of the major quality
factors in models in general is consistency (Lindland et al. 1994; Krogstie et
al. 1995; Becker et al. 2000). As shown in Chapter 4 consistency is also
highly important in business process reference models.
In another field, much work has been done to improve the process of
transforming natural language requirements into formal models that are more
suitable for system design e.g. (Kang et al. 2002). Natural language is not
ideal for system development because it is prone to being unstructured,
having gaps in information and containing inconsistencies (Ryoo et al. 1999;
Fabbrini et al. 2000).
Combining these two facts, i.e. that business process reference models need
to be consistent and that the use of natural language can easily generate
inconsistencies, it is a small step to conclude that natural language is not the
most suitable language for a business process reference model and hence a
semi-formal language backed with textual descriptions is used.
The use of a graphical language makes the model more accessible by a
wider audience and provides a basis for software design if the users of the
model wish to use it for this purpose.
State
Many descriptions of current practice facilitate the development of standards
particularly with respect to electronic communications or interfaces (for
example the ISO-OSI or other technical telecommunication and IT
standards). The model to be designed describes best practices in IT service
provision. The term ‘best practice’ always produces controversy. It is used
here to indicate that the processes described in the model are the commonly
accepted current viewpoints on what would be the best way to perform the
activities, with none of the common constraints that restrict ‘current’ practices
such as lack of integrated tool support or lack of resources such as
experienced personnel or organisational or even industry cultural legacies.
Other terms that are similar to “best practice” are “good” or “accepted”
practice. The reason that a depiction of best practices as opposed to
Masters Thesis Chris Taylor
QUT Page 131
common practices has been chosen is that there is a general
acknowledgement that current practices in IT service delivery can be
improved. Also as mentioned earlier common practice models are often used
to allow interconnection and communication, neither of which are the goal in
this case.
Focus
As is already apparent, the focus of the reference model to be developed is
on the business side of operations. In particular it looks at selected business
processes of the IT service provider organisation.
The business processes are those processes that provide value for the
customer. They typically involve people and decision making using business
rules or policies and the use of IT applications or other forms of structured
data to support the process.
Level
The models to be developed were Complete, Intermediate and Task. The
reason all the levels of the model were presented stemmed from the original
discussion of reference models for this domain. They either lacked detail or
lacked consistent integration. The model was designed to show how the
levels of detail can or should be integrated. This integration the hopefully
overcomes ITIL's major shortfall of inconsistency.
The intermediate models were the higher level (level 0) models that
described all the processes of an IT service provider. This level forms the
basis from which process improvement targeting could be achieved, as well
as providing an overview of the organisation. At this level, organisational
units and process owners could be attached to processes allowing tactical
decisions to be made.
The Intermediate level models (Level 1) describe the basic steps of each of
the processes while the Task level models (Levels 2+) describe the detailed
breakdown of the processes. They provide detailed guidance on how the
processes should be performed, what systems would be employed and the
Masters Thesis Chris Taylor
QUT Page 132
description of the interactions between the organisational units and
individuals. At another lower level the technical steps performed by the
systems are also explained. The model was designed to depict this level of
detail to facilitate the analysis of IT support for the process and to provide
guidance to individual employees.
The hierarchy of the model is shown in Figure 47.
Figure 47: Interaction of Levels of Abstraction
Functional Area
The model can be seen as either a functional area specific model in IT or as
an economic activity specific model for IT outsourcing. The model is
classified in this discussion as the economic activity specific model, therefore
making it a Function Area of the model “enterprise”.
The model’s core is the IT service provision core processes, and, while the
model recognises the existence of other activities such as strategic
management or the support processes, it does not examine them in any
depth or detail.
Theoretically, the model could also be applied to in-house IT service
providers. In fact due to the trend toward more accountability in IT in-
sourcing, many in-sourced IT providers are mirroring the processes of the
1 2 3
2.1 2.2 2.3 2.4
2.2.1 2.2.2 2.2.3
Incr
easi
ng L
evel
of D
etai
l
Level 0 Complete
Level 1 Intermediate
Level 2+ Task
…
…
…
…
Masters Thesis Chris Taylor
QUT Page 133
outsourcer (Willcocks et al. 1997). This division between the internal ITSP
and its host organisation may not be as advantageous when part of the IT
being provided is actually the core business of the host organisation.
Economic Activity
The model was developed as a depiction of an IT service provider
organisation. As such the industry would be IT, making the model a specific
economic activity model. As discussed above however, the models could
easily be used, in an internal service provision situation (an IT “shop” or
department). This would be particularly true in an organisation with a
centralised IT environment or one that used internal billing as a method of
management and control.
There were several reasons this domain was chosen.
Firstly, existing literature, particularly eTOM and ITIL, covered the domain but
did so in a way that was unsatisfactory for the purposes described above
(see 2.8 ITSP Business Reference Models for this discussion and 6.2.1 for
the purpose of this model). However, these models provide material for the
design phase of the model which can offset the lack of knowledgeable
resources available to the research.
IT service provision, in contrast to many other functional areas is relatively
young, with IT only being used extensively and pervasively to support
businesses relatively recently5. Hence, there is an opportunity for greater
improvements by formalising the process than there is by focusing on
established areas such as accounting, logistics or manufacture etc. This
combined with the established use of modelling to design IT systems makes
the application of reference modelling to the delivery of IT services an
academic and practical challenge.
5 The first commercial electronic calculator was released in the 1960’s, ERPs in the 1980’s, compared to accounting which has been used for over 2000 years (Williams 1997).
Masters Thesis Chris Taylor
QUT Page 134
Tool Support
As described previously the language of the model is eEPC. The decision to
use this language was partly due to the tool support available for the
development and distribution of models developed in this language. Having
tool support was one of the quality factors which were mentioned in
Chapter 4.
The tool package, produced by IDS-Scheer, has been rated “leader” in
Business Process Tools by Gartner (Anonymous 2002). The ARIS Toolset
allows a consistent approach to modelling. It does this in several ways,
mostly by relying on the database backend of the package to deliver
consistent naming and usage of defined objects. Consistency is one of the
most important aspects of a reference model, as described in Chapter 4
A major reason for the choice of the dedicated modelling tool was the
existence of in-built consistency checking functionality. ARIS Toolset already
came with predefined and accepted depictions of the building blocks of
models in the language. It also allowed for the easy distribution of the
developed models and the accurate interfacing between the models. Another
attraction of this toolset in particular was the incorporation of the different
languages selected in the model development, value-added chains and
eEPC. The author’s previous experience with both the tool and the language
were also influencing factors. The current availability of several other
reference models in the toolset provides a proven track record, for the
development and ultimately use of using ARIS as a tool for reference
modelling.
Readiness for Use
The model is a single depiction type model. The model makes several
assumptions about best practice that could vary with the specific situation in
which IT services were being provided. These assumptions are discussed in
Chapter 8. With greater resources variants could be added, depicting a
different process for the different assumptions, for example the different
processes for external vs. internal service providers.
Masters Thesis Chris Taylor
QUT Page 135
Other Characteristics
Characteristics that are to do with the distribution and commercialisation of
the reference model have been excluded because they are not relevant to
this research project. In Table 18, the characteristics that have been chosen
for this model are presented. Due to time and resource constraints no
extended content is provided, however it is expected that the ITIL books
would provide substantial extended content. There are no plans at this stage
to make it a living reference model although there is no reason, other than
the requirement for resources, that this could not occur.
The characteristics of the derived reference model for ITSP for this research
is summarised in Table 18.
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
State Common Practice Best Practice
Focus Business Technical Application
Level of Detail Complete Intermediate Task
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer Supplied Third Party Supplied Public Domain
Readiness for use Single depiction Contains variants Abstract
Table 18: Summary of ITSP Reference Model Characteristics
6.2.3 Modelling Conventions
The modelling conventions have been modified by the input from the focus
groups particularly in relation to the scope of the model. The final modelling
conventions are presented in this section.
Masters Thesis Chris Taylor
QUT Page 136
Model Scope
Using the taxonomy of IT services presented in 2.7.2 and the lifecycle of IT
services presented in 2.7.3 the model is positioned as depicted in Figure 48.
Figure 48: Scope of the ITSP Reference Model
As shown in the Figure 48 (derived in Chapter 2), the model is focusing on
the “Delivery and Support” or the operations phase of the IT Service
Provision lifecycle and on “Service Delivery” and “Infrastructure
Management” in terms of the market taxonomy. This simplified lifecycle
model does not differentiate between the first cycle of service support, which
would include the initial planning and setting up of an entire IT system,
versus the progressive planning and implementation of single applications,
which could happen quite frequently. The model developed acknowledges
that even within the “Support” phase, new applications and systems is
planned and implemented. The model specifically excludes the development
of a new application, because other models have dealt with this area and this
is not the focus of this research. It does however include the processes that
Plan
Implement
Support
Retire
Solution Delivery
Service Delivery
Infrastructure Management
Network Management
a) IT Services Taxonomy
b) IT Services Lifecycle
Scope of Model
Masters Thesis Chris Taylor
QUT Page 137
take the developed application and release it into the production
environment.
The two market layers of Service Delivery and Infrastructure Management
were chosen because the feedback from the initial focus groups and the
review of the literature suggested that large IT outsourcing vendors operate
in these two layers.
Most IT outsourcers offer services in the solutions delivery domain, but
feedback from the focus groups indicates that while these products and
services are often offered they are very rarely utilised.
Modelling Methods and Tool
The modelling package ARIS Toolset was chosen for the design of the
reference model. The Toolset offers a range of modelling techniques one of
which is the Value Added Chain Diagram, used for the mid-level modelling,
and extended Event-Driven Process chains (eEPC), which is the chosen
method the modelling language at the lower levels.
eEPC – Level 2+
Event driven process chains show the chronological-logical procedure of a
process (Anonymous 2002 p4-103). This process modelling language is
driven by events. Events are defined as “the existence of a state”. The events
trigger and are created by functions. A function is a unit of work.
Object Types
The following object types and their descriptions have been adapted from the
ARIS Methods Manual (Anonymous 2002).
Masters Thesis Chris Taylor
QUT Page 138
Symbol Object Type Name Description
Event Events trigger functions and are the results of functions.
Function A function is a technical task or activity performed on an object
IT Function IT Functions are those functions that are
completed automatically by the IT application.
Process Interface
A process interface indicates from which process the related event has been created, or which process the event
triggers.
Rules X-OR AND OR
The rules describe how the events and functions can be related.
The X-OR means that one and only one input/output is possible, the AND that all
the inputs or outputs must be true, and the OR states that any combination may be
possible.
Person Type
A person type represents the typification of individual persons which have the same
characteristics. These characteristics may refer to similar authorizations and
responsibilities, for example.
Application System Type
The application system type is representative of a related group of IT
systems that have the same functionalities.
Table 19: Object Types for eEPCs
Naming Conventions
Events are generally named following the format of “noun” “verb” e.g.
“Incident Recorded”. Functions are named “verb noun” e.g. “Detect Incident”.
Maintained Attributes
The maintained attributes of the objects are shown in Table 20.
Masters Thesis Chris Taylor
QUT Page 139
Maintained Attribute Information contained Descriptions/Definitions Gives a general description of the object
Source This shows from where the ideas for the particular part of the model originated.
Remark/Example Gives examples or further explanatory information Table 20: Maintained attributes in eEPCs
Connection Types
The connection types in the eEPC’s are summarised in Table 21.
Connection Type Description Leads to the function produces the event activates The event triggers the function
Is evaluated by The rule decides the process flow
Contributes to An organisational unit that is involved in but does not execute a function
executes The organisational unit that executes the function Table 21: eEPC connection types
Model Layout Convention
The eEPCs have been produced to show the “error-free” path as close to the
centre as possible, flowing down the page. An error-free path is the path that
an ideal process instance should take. Error or exception conditions are
shown as a branch out form this error-free path. An example of this is shown
in Figure 49. At branching points that do not have a clear error/non-error
situation the paths have been split symmetrically around the centre.
Figure 49: Example of error/non-error process split depictions
Configuration details correct
Check configuration
details
Configuration details
incorrect
X
Masters Thesis Chris Taylor
QUT Page 140
Omitting Trivial Events
After feedback from the focus group, in which an eEPC strictly adhering to
the event-function-event ordering of objects was shown, the trivial events
were seen as adding un-necessary clutter to the model. This provides an
example of the results of case studies presented in Chapter 4 that state that
the language used must be easy to understand. The omission of trivial
events for the eEPC technique is also discussed by Scheer (1998) and is
facilitated in the Diagram Explorer part of SAP’s ValueSAP.
Therefore all trivial events have been removed from the models. An event is
deemed to be trivial if all of the following conditions hold true:
• the event occurs every time after the execution of the function
• the event does not mark the start or end of a sub-process or the start
or end of a process or process step as defined by the value chain
An example is presented in Figure 50.
Figure 50: Example showing removal of trivial event
Combining Optional and Mandatory Paths
In the situation where the process continues and also may trigger another
process path an OR operator is used. The mandatory path is shown centred,
while the optional path is shown as a branch as in the example in Figure
Classify Incident
Accept Incident
Classify Incident
Accept Incident
Configuration details correct
Event removed
Masters Thesis Chris Taylor
QUT Page 141
51(a), where the Service Level Agreement is assigned regardless of the
outcome of the threshold check, and the threshold check may produce the
event “Threshold exceeded”. Alternatives of modelling this situation in a
strictly syntactically correct method are shown in Figure 51(b)I, ii and iii.
Figure 51: Combining Mandatory and Optional process paths
(a) Modelling convention
(b)i Formally Correct Representation
(b)ii Formally Correct Representation
CheckThreshold
Thresholdexceeded
Assign ServiceLevel Agreement
RecordSymptoms
RecordSymptoms
CheckThreshold
Thresholdexceeded
Threshold notexceeded
Assign ServiceLevel Agreement
Notify ProblemManagement
CheckThreshold
Thresholdexceeded
Assign ServiceLevel Agreement
Threshold notexceeded
RecordSymptoms
(b)iii Formally Correct Representation
Threshold Checked
CheckThreshold
Thresholdexceeded
Assign ServiceLevel Agreement
Threshold notexceeded
RecordSymptoms
Masters Thesis Chris Taylor
QUT Page 142
The decision to use the slightly less formally correct option (a) was to make
the models less cluttered and to make the language easier to understand.
The first syntactically correct version (b)i involves a consecutive combination
of two types of connectors that requires reasonable understanding of the
modelling technique to interpret. The second correct form of modelling the
situation is shown in (b)ii, it involves 50% more objects than the proposed
form. The third syntactically correct depiction (b)iii, shows a slightly different
semantics, indicating that as soon as the “Record Symptoms” has occurred
the Service Level Agreement can be assigned. All of the “correct” depictions
contain extra objects, and are potentially more difficult to understand quickly
and easily, which was one of the indicators of quality identified in Chapter 4.
Process View - Value Added Chain Diagram (VAC) – Level 1
The Value Added Chain Diagram (VAC) originally proposed by Porter (1980)
is a simple form of modelling that shows the successive functions that make
up a process. It usually shows a higher level of abstraction than the eEPC
notation, depicting about 5-10 major steps in the process. It can also be used
to show levels of the functional hierarchy. In the reference model it is used to
depict the level 1 process of Incident Management. Incident Management
has a logical sequential flow of activities that makes the use of a value chain
appropriate. Other ITSP processes may not have this logical flow, and may
be more accurately depicted in other modelling techniques such as a function
tree. Because this research focused on the Incident Management process,
the other modelling techniques are not explored.
Object Types
There is only one object in the value chain; the function object type shown in
Table 22.
Masters Thesis Chris Taylor
QUT Page 143
Symbol Object Type Name Description
Function Functions in the VAC indicate high level steps of a process
Table 22: Object Types for VAC
Connection Types
There is only one connection type in the VAC described in Table 23.
Connection Type Description Is predecessor of Means that the value chain step is performed before the next step
Table 23: Connection Types for VAC
Maintained Attributes
The maintained attributes of the VACs are shown in Table 24.
Maintained Attribute Information contained Descriptions/Definitions Gives a general description of the object
Table 24: Maintained Attributes for VAC
Model Layout Convention
Processes that have a logical sequential nature are shown from left to right.
Functions that do not have a logical place in the process are shown top to
bottom as in Figure 52.
Figure 52: Example of layout of non sequential VAC steps
Business Process Framework – Level 0
The business process framework is the highest level in the model. It is a free
form model, with no underlying modelling technique. The framework should
contain all the functions of the IT service provider.
1 2 3
A
Masters Thesis Chris Taylor
QUT Page 144
It is divided up into process groupings similar to those used in eTOM. eTOM
uses the concept of the supply chain with customers at one end of the model
and suppliers at the other. In the other dimension the model uses a concept
of lifecycle time, which is also related to the strategic vs. operational
continuum. These two dimensions are reused in the ITSP reference model at
the process groupings and, to a lesser extent, the level 0 view.
Figure 53: Dimensions in the ITSP Level 0 view
Model Hierarchy Structure
The models hierarchical structure is based on that used in eTOM and SCOR.
That is, a top level model which, under each object, contains increasingly
detailed models. The top layer is termed Level 0, and each increasing level of
detail increases the number of the level. Figure 54 depicts this.
1 2 3
2.1 2.2 2.3 2.4
2.2.1 2.2.2 2.2.3
Incr
easi
ng L
evel
of D
etai
l
Level 0 Complete
Level 1 Intermediate
Level 2+ Task
…
…
…
…
Strategic processes
Operational Processes
Customer facing
Processes
Supplier facing
Processes
Masters Thesis Chris Taylor
QUT Page 145
Figure 54: Model Hierarchy Structure
In the models developed there is no sharing, i.e. the hierarchical relationships
were strictly 1:m. However, with the further development of the model it is
conceivable that sub-models could be re-used in various high-level models.
When this re-use is appropriate the modelling conventions should be
changed slightly to incorporate the structure of these shared models.
6.3 Design and Validate High-Level Process Framework
The loop between the design and validation step occurred 2 times for the
high level model (Level 0), by the end of the second validation the
participants and researchers were satisfied that a level of economic efficiency
of design had been reached. This section combines the Design and
Validation phases for the Level 0 model.
6.3.1 Design Philosophy
Due to the purpose of the model, that is to provide a best practice reference
model to be used for Business Process Management, which can also be
applied to the evaluation of a software package to support ITSP, the main
design philosophy adopted by the research team is a combination of Blue
Sky and Best Practice by Composition design, much of the material is drawn
from ITIL, due to the lack of other information sources containing the detail
required for a Task level model.
The two design philosophies used in conjunction, allow good existing ideas,
usually point solutions, to be combined in a way that integrates new with
existing ideas. This applies to both the content as well as for the ways of
expressing it.
A graphical way of representing the combination of the two design
philosophies is outlined in Figure 55.
Masters Thesis Chris Taylor
QUT Page 146
Figure 55: Combining Blue Sky and Best Practice Design Philosophies
6.3.2 Identify Information Entities
Two main groups of information entities were identified that would be useful
to the ITSP reference model, those from literature and existing models, and
industry IT service providers.
Review of Existing Models and Literature
As presented in the literature review in Chapter 2, there are a number of
reference models for related domains. Of these models two (i.e. MoF and
Corbit) reference ITIL as a major source of input. ITIL is one of the most
popular reference models for IT service management (Morin 1999; Duffy
2001; Dubie 2002; Kara 2002) and one of the only to offer detailed process
descriptions in the public domain. It lacks a comprehensive model at the
higher levels and in the integration of the ITIL processes, as well as a method
of accurately depicting the content. This is where other model identified in
Chapter 2 are helpful. Hence at the lower levels the input of ITIL is most
relevant in the reference model design although different ideas where drawn
from other sources as explained throughout the text. The other information
sources included help desk application manuals, other models discussed in
Chapter 2 and theoretical modelling precedents drawn from eTOM
Reference Model
Sub-entities created (Blue
Sky)
…
Re-used sub-entities (Best Practice point
solutions)
Existing Information Entities
Masters Thesis Chris Taylor
QUT Page 147
especially. The individual sources are discussed in detail in section 6.3.3 of
this chapter where they are related to the final reference model.
One of the most important sources of input came from experienced
managers currently providing ITSP to paying customers. This input was
gathered through focus groups.
In order to provide validity to the models and to provide feedback on the
modelling techniques, focus groups were designed to provide the validation
step. By gaining the feedback from organisations that are currently providing
IT services, the quality aspects such as providing a semantically correct
model and easily understood modelling language could be improved. The
focus group sessions also provided an atmosphere where the designers (the
research team) and a representative sample of the potential users could
engage in discussion and provide the feedback loop from important
stakeholders. This was identified in Chapter 4 as one of the attributes that
contributed to quality in a reference model.
Focus Groups
This section starts with the focus group recruitment, a brief overview about
the organisations involved, the focus group settings and the data collection
techniques.
Focus Group Recruitment
"A critical aspect of conducting focus groups is to specify the inclusion and
exclusion criteria for participants" (Sofaer et al. 2001). Possible participants
for this focus group were sourced from large application service provision
companies in Australia. Organisations that provided large IT service delivery
to outside organisations or, to their own companies, in-house providers were
targeting for participation. Particular attention was given to those with suitable
staff based in Brisbane. The result provided commercial and government
outsourcing organisations.
Participants were sourced initially by two methods;
1. Known contacts of the research team
Masters Thesis Chris Taylor
QUT Page 148
2. Contacted from research into commercial and government organisations
conducting large IT service provision service delivery operations
Initial contact with possible participants was by email and if interest was
expressed then a face to face meeting was scheduled. At this first meeting
the research team provided information on the scope of the project and the
benefits of participation. (The combined project was the ‘Process oriented
administration of enterprise systems’ 6 and not only the research project
described in this thesis.) In addition we described the possible solutions to
problems with confidentiality.
Participants
The following participants contributed to the research by participation in the
focus groups; IBM Global Services Australia, Computer Science Corporation,
EDS Consulting, Mincom, Hitachi Data Systems, Deloitte Touche Tohmatsu,
Citec and REALTECH AG. IBM Global Services Australia, EDS, CSC and
Citec on their own are considered the clear dominant players within the
Australian market, claiming nearly 80% of the annual IS Outsourcing
revenues based on figures from (Benson 2002). The participants in the
research were in management positions and had job titles such as Strategic
Service Manager or Service Support Manager. See Appendix E for
descriptions of the organisations involved.
Focus Group Setting
The focus groups were conducted in a meeting room at QUT’s Centre for IT
Innovation (CITI) in a central location in Brisbane CBD. Refreshments and
lunch were provided during the focus group sessions. A PowerPoint
presentation was used to orient the focus group and participants were
randomly seated around a large table in a board room type setting.
A total of 3 sessions were held at a spacing of about 3 weeks each. The first
focus group’s topic was the high-level model, the second focus group was
6 This project was funded under the ARC Linkage program operated by the Australian Research Council with the industry partner REALTECH.
Masters Thesis Chris Taylor
QUT Page 149
going over the changes made to the high-level model and starting the level 2
model for Incident Management, and the last session was on the Incident
Management process eEPC.
The sessions had 4-7 people in attendance and were all around 2.5 hours.
Data Collection
Data collection in focus group research used audio-taping, note taking and
the changes to the model made by the participants. The approach taken for
this focus group was to use two audio-taping machines, a dedicated note
taker and timekeeper in addition to the moderator. Immediately following the
session the note taker, timekeeper and moderator composed their main
thoughts and issues. All audio tapes from the focus group were transcribed
and checked and this information was sanitised to remove names or other
identifying information. This approach is consistent with those recommended
by (Morgan 1988; Saulnier 2000b; Fern 2001). These comments although
important were only secondary data. The suggested changes to the models
were written by participants directly onto large printouts of the model which
had been supplied at the beginning of the focus group. Proposed changes to
the models were discussed at length by the focus group, the comments both
written (on the models) and oral and the marked printouts of the models
constituted the data collection from the focus groups.
6.3.3 Build
The steps of relating information entities that described the high-level
processes of ITSP and rating these entities are incorporated into the
following discussion. The relation and rating of the identified entities was
derived from the literature review.
The scope of the model is IT service provision. This scope is quite broad and
during the focus groups the scope was refined to more closely match the
industry realities. The purpose in this case is to provide a logical structure to
classify, present and perceptualise about. It is the starting point for a user to
enter the model, providing guidance about the content and where to find the
content they are seeking, while not overwhelming the user with detail.
Masters Thesis Chris Taylor
QUT Page 150
The development of this high-level model focuses on the core activities of the
application service provider organisation, and the support and strategic
processes are left underdeveloped. The reason behind this is that other
reference models exist which cover these areas, and the unique contribution
of this research is in the area of IT services. Hence modelling how to perform
related non-core processes for example training or financial management
would not create significant value and is largely regarded as being out of
scope.
Most of the model in terms of constructs and content was derived from ITIL
and eTOM, two existing industry reference models. These models were
chosen due to their large contributor base and industry acceptance (Dubie
2002; Kara 2002), and their relation to the domain in question.
The modelling was conducted using a top-down methodology. There were
several reasons for this; the main reason was the criticism of ITIL for not
having a consistent approach and format. Also other major reference models
appear to have been developed using a top-down approach and a higher-
level model is more widely useful than a lower level model. This research
was not aimed at developing all the lower level models and therefore the
most useful contribution in light of the constraints of the this research would
be a higher-level model with selected lower-level models.
The credited original (and certainly most popular) method aimed at modelling
the processes of an organisational entity is Porter’s Value Chain (Porter
1985). In the model, the “Primary Activities” (or processes) are the ‘value-
add’ing activities (i.e. In-bound Logistics, Operations, Outbound Logistics,
Marketing and Sales, and Service). These processes are enabled by the
“Support Processes” (Human Resource Management, Technology
Development and Procurement). The developed diagram (or process model)
is depicted in Figure 56.
Masters Thesis Chris Taylor
QUT Page 151
Figure 56: Porter's Value Chain
Porter’s value chain was developed when the manufacturing of physical
goods dominated thinking about the firm and industries. For service
industries, particularly of an electronic nature, notions of physical logistics
and production are less important.
At the highest level Porters value chain can be depicted as in Figure 57.
Figure 57: Highest level of Porter's Value Chain
Moving from left to right the transformation of goods through the Primary
activities (and the additional value) can be tracked. Perpendicular to these
‘value-adding’ processes are the support activities. These enable the Primary
activities. Using this level of abstraction Porter’s value chain can be applied
for service industries.
Support Activities
Primary Activities
In
boun
d Lo
gist
ics
O
pera
tions
M
arke
ting
and
Sale
s
S
ervi
ces
O
utbo
und
Logi
stic
s
Procurement
Human Resource Management
Technology Development
Infrastructure
Masters Thesis Chris Taylor
QUT Page 152
The IT services taxonomy presented in Chapter 2 forms the basis for the first
model. To match Porter’s Value Chain it is turned 90 degrees and has the
enabling processes at the bottom.
The first version of the IT service provision business process reference
model is depicted in Figure 58.
Figure 58: ITSP first version
The model is essentially the top three levels of the IT taxonomy presented in
Chapter 2.
The term “solution delivery”, was changed to Complementary to capture
expert services with a larger scope that Tao’s Solution Delivery, which was
essentially the packaging of the underlying services to present a single face
to the customer (Tao 2001). This includes services designed to supplement
or improve the other layers of the IT service and the services that reached
into the consumer organisation such as business process management or
strategic IT consultancy. The Complementary process encompassed
consulting services about how to, when to and which applications and
systems to use.
Product definition is added to show that the delivery of all IT services should
be preceded by a negotiation and definition of the services to be provided.
Masters Thesis Chris Taylor
QUT Page 153
Product definition involves the negotiation of the service scope and detailing
of the specific service to be delivered. Application Hosting is the technical
housing and maintenance of the applications and includes the infrastructure
from which the service was provided, subsuming the infrastructure
management. Using the platform provided by Application Hosting, Service
Delivery, involves the maintenance of the system in order to provide the
specific services as promised. It also includes Service Support which
involves dealing with problems or Incidents that occur in the system from a
service degradation perspective, that is, as perceived by the client.
During this development time eTOM v2.5 was released (Anonymous 2001).
Its higher-level frameworks, depicted the supply with an “Enterprise
Management” process at the bottom, which is very similar to the Enabling
process (it includes lower level processes such as HR, Financial and
Technology management).
The Application Hosting and Service Delivery are very similar to the two
processes “Resource Management and Operations” (including Application,
Computing and Network) and “Service Management and Operations”.
Effectively it was similar to the first ITSP reference model although rotated 90
degrees left as shown in Figure 59.
Masters Thesis Chris Taylor
QUT Page 154
Figure 59: eTOM (v2.5) Level 1 processes
One aspect present in the eTOM model that is missing from the original ITSP
model is the strategy section. Along the left hand side of the eTOM model
lies the Strategy, Infrastructure and Product Processes. The idea of long term
strategic business and product planning was incorporated into the ITSP
model as shown in Figure 60.
Masters Thesis Chris Taylor
QUT Page 155
Figure 60: ITSP Second Version
The names and structure of the high-level processes from eTOM were not
used for several reasons. eTOM was developed for the ICT market, which, at
least in part, operates in a mass consumer market. Operating in this market
necessitates a different type and emphasis on marketing. Due to the size and
relatively long-term commitment of an IT service provision agreement,
marketing (in the traditional mass marketing concept) is seen as of lesser
importance to the target audience of the developed model. Hence, eTOM’s
“Marketing, Product Customer” process was not included in the high level
model. eTOM’s supplier management is still an important aspect in ITSP but
not as important as it is in the telecommunications industry where real time
interactions with many other telecommunications companies are an essential
part of business. ITSPs usually deal with one telecommunications supplier in
a much simpler relationship. They are essentially outsourcing the
management of complex communications relationships to the
telecommunications supplier, making supplier management a support rather
than core process and hence are not shown on the high level model.
The process naming at the lower levels was also influenced by the recently
recompiled titles and descriptions of the ITIL books, i.e. Service Support,
Service Delivery, Security, ICT infrastructure management and Application
Masters Thesis Chris Taylor
QUT Page 156
Management. ITIL is more specific to the domain in which the model is
based.
The level 1 view was developed by adding more detail and was the first
presented to the focus group for validation (Figure 61).
Figure 61: Draft of the ITSP Model as presented to the focus group
6.3.4 Validate
The first draft of the ITSP reference model was presented to the focus group
on 18th of July 2002 and in the beginning of the next focus group on the 15th
August 2002. The first 30mins was spent orienting the participants, and
providing definitions of “reference model” and the proposed structure for the
reference model. The scope of the model was also discussed, originally the
scope statement was given as a model of “the systems and activities needed
to provide a business end-user with access to an ERP in order to derive
business benefit”.
The two views of the model were presented (the highest level process
groupings and the level 1 view as shown in Figure 61).
The initial topic that was raised by the participants was confusion about the
subject of the model. The participants suggested that, having discussed the
Masters Thesis Chris Taylor
QUT Page 157
scope of the model further, by using the taxonomy of IT services and the IT
lifecycle, that the Product Definition grouping should be removed, as these
processes were only conducted at the beginning of a relationship, particularly
the contract negotiation. When prompted however, certain aspects of the
Product Definition were part of the operation of providing IT services to a
customer. These processes the definition of service (with process of defining
the service scope and service levels) were seen as an essential on-going
maintenance of matching the client needs to the services provided, and were
done on a regular basis, not a once off process at the beginning of a
relationship.
As a result of their comments the Product Definition was replaced by the
“Service Definition” group, which was moved from the left hand side of the
model, which was made the model appear more a lifecycle, to the top in
accordance with the top down strategic to operational flow of the model
similar to eTOMs.
The participants also indicated the arrows indicated that the model was itself
a lifecycle, as opposed to a process framework. The arrows were kept
however (later to be removed in testing where they were again identified as
unnecessary), because they were perceived by the research team to be
integral to the model, especially in relation to Porter’s value chain.
The participants also argued the model as it was presented and described
could represent all IT services, not just ERP services. They also indicated
that ERP services, in their experience, were often not provided in isolation to
other IT services. Hence the name of the model was changed from ERP to IT
services.
The Complementary services were discussed next, the focus group
suggested that, although most ITSP’s offered such services, that the uptake
of these services was so low that it was not relevant or within the scope of IT
services. These activities were also seen as more strategic issues that were
not within the more operational focus of the model.
Masters Thesis Chris Taylor
QUT Page 158
The effect of these comments was the removal of the complementary
services group from the model.
The hosting group was renamed Service Infrastructure Hosting to distinguish
it from web page hosting, and the processes in this grouping were seen as
correct and relevant.
At the suggestion of the moderator, a distinction in the two groups of the
customer was added, to aid discussion. The focus group members
suggested that there were in fact two distinct customer groups, the clients,
i.e. the managers of the customer organisations who controlled the budgets
to be spent on outsourced IT and the user group, the actual business
operators who consumed the IT services.
After identifying these two groups in the consuming organisation, the model
was re-arranged, to show which processes targeting which customer group.
In keeping with the top to bottom flow of the model from strategic to more
operational, the clients, those controlling the budget, were situated at the top
of the model. In keeping with Porter’s value chain idea, to the right of the
model where all the customer facing processes were placed (see Figure 62).
After discussing the re-arrangement, several processes that were not
discussed in the draft model were identified including Reporting and Billing.
These processes were grouped into a new process grouping called
Customer Relationship Management.
Feedback from the participants indicated that the service support functions,
were an important part of IT service provision, and deserved the same
importance as the other processes already mentioned. Hence the “Service
Support” process group was added and the support processes were added
within this group, drawing heavily on ITIL.
These changes resulted in the model shown in Figure 62.
Masters Thesis Chris Taylor
QUT Page 159
Figure 62: Reference Model showing "Customer" object on right
6.4 Select Low-Level Process
After a first focus group and Delphi study that formed part of Craig Huxley’s
(Huxley 2003) work on identifying the processes that organisations should
focus improvement activities upon, the processes for modelling at the lower
level was selected. This selection was based partly on the results of the
focus group and Delphi study into the critical processes of an IT Service
Provider. The other consideration was the availability of material on the topic.
The topic chosen for modelling was Incident Management. Incident
Management is a relatively structured process. Structured processes are
more suited to the chosen modelling technique (Scheer 1998 p556).
Processes that are poorly structured may need alterations to the modelling
conventions or even the use of other modelling techniques.
6.5 Design and Validate the Low-Level Model
The steps of defining the design philosophy and identifying information
entities have already been discussed, leaving the steps of building and
validating the lower level models.
At the lower levels the ITIL books provided the most detailed guidance. None
of the other best practice frameworks had a similar low level advice on how
Users
Clients
Customer
Masters Thesis Chris Taylor
QUT Page 160
to provide IT services. Hence, ITIL was the starting point for the models at
the lower level. ITIL is a natural language that contains graphic depictions, an
example is shown in Figure 63 (CCTA 2000 section 8.3)
Figure 63: Example of ITIL text and diagrams
These graphical representations are ad hoc in design and used through out
the mostly text explanation. Often there is no real link between the text and
the models, for example not all the objects in the graphics are explained in
the text, none of the connections, or symbols are explained etc.
The conversion from textual descriptions into the semi-formal language of
eEPC involves a degree of interpretation. Considerable time was spent
reading and understanding the various parts of the ITIL chapters firstly to
gain an understanding of the material then to determine how the content
should be depicted. An initial eEPC was designed showing the end-to-end
Incident management process and then presented to the next focus group.
An example of the type of change made to the model included the Incident
logging. The focus group members agreed that all calls to the Service Desk
should be recorded in the Service Desk software, instead of only Incidents
requiring resolution as was depicted in the model.
Masters Thesis Chris Taylor
QUT Page 161
Feedback from the focus group was that the models were too detailed to
provide an overview about the process. In particular the inclusion of the trivial
events was seen as unnecessary duplication. The focus group members
indicated that, while the detailed depiction should not be lost, the model
needed an intermediate layer that could communicated the end to end
process in a few simple steps.
With this feedback, a second model was produced using the value-chain
technique. The level of detail of this value chain was guided by one of the
internal models provided to the research team by one of the focus group
participants. The purpose of this internal model was slightly different to the
ITSP reference model in that it was designed for use with clients, so it
concentrated on highlighting the interactions with the client during the
process as opposed to providing guidance for internal use, so at some
stages the level of detail in the company specific model was excluded in the
value chain developed for the reference model. This combined with a
graphical model from ITIL formed the basis of the level 1 value chain for
Incident Management.
Although ITIL formed the basis of the reference model, several parts of the
developed model were influenced by other sources. Two tangible and
specific examples of this are the use of application supported service
requests and the use of “whiteboards”. The application support for Incident
logging came from the SAP Solution Manager. The Solution Manager
interfaced to a user-transparent interface built into the application (in this
case R/3) which allowed the user to submit an Incident report from within the
application. This use of application driven Incident logging and submission is
also seen today in Microsoft’s “Office” package. Essentially, by submitting a
help request or event record through the functionality supported from within
an application, important information can be automatically captured to help
classify, diagnose and resolve the Incident. For example, by using the
reporting functionality that is interfaced with SAP R/3, the transaction code,
license, release, database version, module etc. as well as the user ID can all
be gathered automatically by the system, with the user adding any free text
comments and then submitting the Incident to the Service Desk, without
Masters Thesis Chris Taylor
QUT Page 162
leaving the application,. This type of “smart reporting” can automate the
some of the steps of Incident management.
The “whiteboard” concept comes from reviewing the documentation of the
REALTECH Helpdesk software. ITIL suggests that every Incident that has
been reported should be acted upon to find a resolution and restore services.
The helpdesk software from REALTECH realises that many Incidents may be
reported that apparently have the same root cause, and that resolving the
cause will restore all the services about which the Incidents have been
reported. The most commonly cited example is multiple reports of email not
working, suggesting that the email server is down. Resolving a single
Incident by restoring the email server will resolve all of the email related
Incidents. In large support and service centres hundreds of calls could be
received about such an Incident, it would not be efficient to investigate and
attempt a resolution for each of these calls. The “whiteboard” concept by
REALTECH allows Incidents to be attached to one another, effectively the
newly related Incident is parked while the original report is investigated and
resolved, which may either restore the services for the new call, or provide a
work-around which can be implemented (but not investigated) many times.
These are two examples of non-ITIL concepts that were included in the
models.
Masters Thesis Chris Taylor
QUT Page 163
Chapter 7: Reference Model Testing
7.1 Introduction
This chapter describes the reference model testing, of both the high level and
lower level models. The testing is an important step in the design of a
reference model, as it ensures that the reference model is of sufficient
semantic and pragmatic quality.
The semantic quality assurance is performed by checking that the processes
contained within it can handle common real life situations that, for whatever
reason, were not apparent during the build and validate steps.
The pragmatic quality is tested by allowing those not involved with the
development of the model the opportunity to use the reference model and
make comments or changes. Examining how the model was used can also
provide information about whether the model was interpreted to mean what it
was intended to mean.
The high level model was tested and improved through use in Craig Huxley’s
work (Huxley 2003), using the Level 0 model as a reference model for the
process identification in order to test the developed process targeting method
based.
Because of the nature of the testing and the amount of description, the
reference model testing has been structured into its own chapter.
The lower level models (levels 1 and 2) were tested and improved through 2
case studies, where the reference model developed in Chapter 6 was used
for business process re-engineering projects.
Both of these testing scenarios are described here.
7.2 Test High-Level
The high level model was used as a process identification tool for the
process targeting phase by Craig Huxley, a member of the research team for
the parent research project. This constituted the testing phase for the high-
Masters Thesis Chris Taylor
QUT Page 164
level model. As discussed in Chapter 6 the high level model was designed
and validated using focus groups. Huxley’s work involved case studies were
imbedded in action learning cycles with different companies involved with
each of these cycles (Huxley 2003).
For the purposes of this study only the initial stages of the Huxley’s case
studies are relevant. The first stages mainly the “Define Scope” in step 2 of
Huxley’s ten steps, is relevant because this is were the reference model was
applied.
The process targeting methodology’s ten steps, which was the output of
Huxley’s work, are shown in Figure 64 and are related to the BPM Lifecycle.
Figure 64: Rosemann’s BPM Lifecycle relationship to Huxley's Target methodology
Huxley’s case studies worked through the 10 steps of the targeting
methodology and the reference model was used in the second step. The
Process Targeting
Processindentification
AS-IS Modelling
Analysis
TO-BE Modelling
Process Implementation
Process Execution
Monitoring/Control
1: Pre Planning
2: Define Scope
3: Assess Dependency
4: Assess Failure rate
5: Develop BSC
6: Assess Impact
7: Calculate Criticality
8: Assess Cost/Benefit
9: Prob. of Success
10: Select Process
Masters Thesis Chris Taylor
QUT Page 165
reference models were then customised to suit the organisation involved and
formed the list of processes from which one or several would be selected.
Three case studies were conducted, however only two of the case studies
used the reference model for the scoping and identification step. The third
case study organisation had already developed a high level list of processes
and therefore did not need a process identification step.
The model was presented on A3 paper in the early stages of the case
studies, and participants (representatives from the organisations involved)
were invited to re-use as much of the model as they felt relevant and
appropriate. The reference model presented to the first organisation is
presented in Figure 65.
Figure 65: High Level model used for first case study (Core processes only)
Consistent with the action research approach used by Huxley, some of the
changes in the first case study were carried over into the second case study.
This included the reference model changes made in the first action research
cycle; hence the second case study used the reference model including the
modifications made in the first case study.
Also, as mentioned in the previous chapter, due to time and resource
constraints the changes to the high level model were not re-validated.
Masters Thesis Chris Taylor
QUT Page 166
At a higher level the notion of “best practice” loses some significance,
because it is mostly about the “what” and it is very difficult to convey the
“how” at this high-level. Therefore the main objective of the testing phase for
the high level is to ensure adequate coverage of the processes an IT Service
Provider undertakes.
7.2.1 Case Study Design
Selection of Participants
Participants were invited from those already involved with the research, that
is, large IT services outsourcing vendors.
Participants for the two case studies were sourced from the focus group
participants or invitees. There was a preference for case study sites with
headquarters located in the city of the research team, though interstate
centres were also considered. Each participant in the focus group and Delphi
study approached regarding their interest in participating in the case study
research.
Two national companies agreed to participate in the case study phase. All
were focused on outsourced services in the IT service delivery industry with
two from the commercial arena and one a government owned commercial
entity with customers from both government and public organisations.
(REALTECH AG, CSC and Citec)
Data Collection
The primary source of data was the modified reference models that were
produced during the process identification activities and the description of the
changes that have been documented in Huxley’s Master’s Thesis (Huxley
2003).
Data Analysis
The modified models were compared to the original high level model.
Differences between the models and the reference model were identified.
Also the pre-existing list was compared to the original reference model. The
differences recorded were the number modifications the existing model,
Masters Thesis Chris Taylor
QUT Page 167
addition of parts of the model, deletion of parts of the model and the degree
of structural changes to the model.
The number of changes was assessed by to recorders, the author and
Dr Christian Probst from the University of Muenster. Dr Probst is experienced
in both process modelling and the domain of IT service provision. The degree
of structural changes to the model was assessed in a subjective manner, with
three degrees of changes, minor, moderate and extensive. The models were
rated individually and then the differences in opinions were discussed and a
concessus reached.
7.2.2 Case Study Results
Case Study 1
The organisation involved with this case study was also involved in the
generation of the high level model through the focus group. The model that
was used for the list of processes is presented in Figure 66, the new
processes have been circled.
Figure 66: Enterprise specific model of first case study
There are 4 additional processes and the layout and structure of the
reference model was modified. “Define Billing” and “Define Reporting” were
Masters Thesis Chris Taylor
QUT Page 168
added to the model to complete the Service Definition processes. The
addition of the “Quality Processes” indicated the need to have operational
quality assurance processes and “Enhancement Management” was added
which was described as the processes that took place that were aimed at
fitting the services offered by the provider to the needs of the consumer.
A structural change was to re-define the scope of “Service Delivery” to
include the older service delivery processes as well as the service support
processes. The reasoning behind this change is that the “service” offered to
the customer consisted of both the old service “delivery” and the service
“support”. The change also provided a neat separation between the back-
office “Service Infrastructure” which was generally a common platform from
which multiple customers were serviced and the customer-specific service
delivery.
A subtle change is the replacement of the arrows with the rectangles. The
case study participants thought the arrows confused the model.
This raises an interesting situation for the derivation of the model, where
single process definitions for the infrastructure backend could be combined
with multiple instances of customer specific processes for the Service
Delivery which could be different for each customer. Such tailoring of front
end processes to individual customers would only make business sense if
the customers were important enough to warrant a custom solution. It has
been noted that these front end processes could be matched to different
service level types (e.g. gold, silver, bronze). It is also conceivable that back-
end infrastructure processes could be individualised per customer or service
level type.
Case Study 2
The modified reference model was presented to the second case study
consistent with the action research methodology employed by Huxley. Two
processes were added to the high-level model, the “Proactive Management”
and the “Site Visit Management”.
Masters Thesis Chris Taylor
QUT Page 169
Figure 67: Enterprise Specific Model for Case study 2
The “Proactive Management” was similar to the “Enhancement Process” in
that it was aimed at bringing new ideas and solutions to the attention of the
customer.
“Site Visit Management” was an acknowledgement that service providers,
may need to maintain and manage on-site resources such as a local Service
Desk.
Also the Change, Enhancement and Release Management processes were
extended into the CRM space to show that these processes should be
closely linked to the customer facing processes.
Another change was the renaming of the Service Delivery to “Remote
Support Services”, which reflected the company specific situation.
One of the perceived shortfalls in the original reference model was the lack of
detailed descriptions and definitions of the processes, which corresponds to
the quality factor of having a defined language and scope
Site visit Management
Service Definition
Define Service ScopeDefine Service Levels
Define Billing
Define Reporting
Service Infrastructure (Hosting)
H/WMgmt
S/WMgmt
AppMgmt
Security/Continuity Mgmt
Service Delivery
Monitor Service Levels
Manage Capacity
Configuration
QA
Qua
lity
Proc
esse
s
Customer Relationship ManagementBilling
Reporting
Rel
ease
Mgm
t
Prob
lem
Mgm
t
Cha
nge
Mgm
t
Inci
dent
Mgm
t
Hel
p de
skM
gmt
Service Support
Enha
ncem
ent
Mgm
t
Proactive Mgmt
Remote Support Services
Site visit Management
Service Definition
Define Service ScopeDefine Service Levels
Define Billing
Define Reporting
Service Infrastructure (Hosting)
H/WMgmt
S/WMgmt
AppMgmt
Security/Continuity Mgmt
H/WMgmt
S/WMgmt
AppMgmt
Security/Continuity Mgmt
Service Delivery
Monitor Service Levels
Manage Capacity
Configuration
QA
Qua
lity
Proc
esse
s
Customer Relationship ManagementBilling
Reporting
Rel
ease
Mgm
t
Prob
lem
Mgm
t
Cha
nge
Mgm
t
Inci
dent
Mgm
t
Hel
p de
skM
gmt
Service Support
Enha
ncem
ent
Mgm
t
Proactive Mgmt
Remote Support Services
Masters Thesis Chris Taylor
QUT Page 170
Results summary
In Table 25 is a summary of the background of the 2 models and the number
and types of differences between the models and the original reference
model.
The table shows whether the reference model was used to develop the
enterprise specific model, whether the organisation involved in the case
study was directly involved in the creation of the reference model, and the
number of changes to the models. The final column shows the degree of
structural change to the model, which captures changes to the layout of the
model.
Model Model Background Changes
Used RM Involved with dev.
# Name changes
# Additions # Deletions Structural
changes Modified Model 1 Y Y 0 4 0 minor
Modified Model 2 Y N 1 6 0 minor
Table 25: Summary of Differences to Reference Model
7.2.3 Limitations and Conclusions
There were several limitations in this phase of the research. Firstly, as
explained in Chapter 6 the changes to the model were not re-validated
through focus groups. Hence the decisions on which were incorporated into
the final model (discussed below) were left entirely to the research team.
A second limitation is that the case studies were designed to test Craig
Huxley’s targeting methodology, not specifically to collect data about
changes to the reference model and hence the model was not used for any
further steps in the Business Process lifecycle.
The third limitation was the use of action research, as the output from the first
was used as input to the second case study. Although this is a strength of the
research, which allowed the refinement of the targeting process, having two
independently modified reference models may have given an interesting
comparison and a more valid insight into which of the changes were
organisation specific and which were needed to improve the reference
model.
Masters Thesis Chris Taylor
QUT Page 171
Another limitation was the buy-in of the organisations. It was clear that the
participant organisations were not attempting to embark on a whole-scale
business process re-engineering effort, the result being that insufficient time
and resources were expended on modifying the high level model thoroughly.
The final limitation is that the changes to the reference model were not
facilitated by an expert, although this may not be a limitation as such,
because the model may eventually be used by non-experts.
The case that had an existing process framework raises an interesting
question in regard to the relationship between process management maturity
in an organisation versus their need for a reference model or need for a
specific type of reference model, and how existing models can be used in
conjunction with reference models. It could be argued that an organisation
with a mature process framework and definitions would not benefit from a
reference model as much as an organisation without a mature enterprise
model. It would certainly be true that the reference model’s application as a
template would be limited, but the other uses for a reference model could still
make it useful to an organisation. One of the most interesting uses for a
reference model in a mature organisation would be the semantic mappings
between the internal processes and that of the reference model. This
semantic mapping could help to communication to other organisations that
would obviously not be using exactly the same models and constructs as one
another. This could help create a common terminology within an industry, as
is the intent with models such as eTOM.
As explained in Chapter 2 it is unreasonable to assume that a reference
model can be used without some degree of customisation. With this in mind,
providing the appropriate reference model was selected for the situation, the
reference model quality for a specific domain could be partly inferred from the
number of changes to the model. This mirrors the “economic efficiency” of
use discussed in Chapter 4, in that a model is high in quality if it is efficient to
use. One of the costs of a model is the required changes to instantise the
model, consuming time and resources.
Masters Thesis Chris Taylor
QUT Page 172
Hence, it is reasonable to conclude that, all other factors being equal, the
fewer changes are needed the higher the model quality. The testing phase is
used to determine the quality of the model and to identify any changes
needed to improve the quality. If there are no changes to the model during
the testing phase then it can be assumed that the model is of reasonable
quality particularly with respect to the economic efficiency of use. From the
results shown above, the models seem to have been reasonably well suited
for the case study organisations with only limited changes, and no deletions.
While conducting the second part of the testing phase, i.e. to increasing the
quality of the model, a distinction needs to be drawn between which of the
changes made during the case studies were enterprise specific, i.e. were
customisations of the reference model during derivation, and which changes
were deficiencies in the original reference model. The deciding factor
governing this decision is whether the changes to the reference models
explained above would improve the reference model as a tool for business
process re-engineering. The following changes were made to the original
model as result of the high level testing. This decision was made by the
coders (Taylor and Probst) based on their knowledge of the specific
situations, the domain in general and using input from other information
entities to judge the generalisability of the enterprise model depictions.
7.2.4 Effect on Reference Model
As a result of the testing phase using the case studies conducted by Huxley
in the Process Identification and Targeting Phases, the following changes
were made to the model.
The “Billing Definition” and “Reporting Definition” were added to the model.
Masters Thesis Chris Taylor
QUT Page 173
Figure 68: Level 0 Processes
The names were changed to align them with the naming conventions and the
Service Desk Management was removed because it wasn’t perceived to be a
real process but rather more of an organisational unit, which was responsible
for some of the processes in Figure 68.
7.3 Test Low-Level 7
The testing of the lower level models was conducted through two case
studies, where the models were provided to consultancy teams engaged with
business process improvement projects with two organisations.
Again the case studies had two purposes, to determine the extent of re-use
of the models, and to improve the models based on the enterprise models
developed and feedback from the model users.
Both of the case studies were in the Incident Management / service desk
domain. Both projects used the business process management lifecycle. The
7 Parts of this chapter section is published at the RefMod2003 Conference in Frankfurt in September 2003, in the paper Taylor and Probst “Business Process Reference Model Languages: Experiences from BPI Projects”.
Service Levels Definition
Reporting Definition
Har
dwar
e M
anag
emen
t
Sof
twar
e M
anag
emen
t
Appl
icat
ion
Man
agem
ent
Security & Continuity Management
Billing Definition
Service Levels Monitoring
Configuration Management
Capacity Management
Service Scope Definition
Rel
ease
Man
agem
ent
Enha
ncem
ent
Man
agem
ent
Billing
Reporting
Site M’ment
Prob
lem
M
’men
t
Inci
dent
M’m
ent
SERVICE DEFINITION
INFRASTRUCTURE SERVICE DELIVERY CUSTOMER RELATIONSHIP
SERVICE SUPPORT
Masters Thesis Chris Taylor
QUT Page 174
first focused on the AS-IS modelling and the process analysis, while the
second project, performed the AS-IS modelling, analysis and produced a TO-
BE model.
7.3.1 Case Study Design
In this phase of the testing the value chain and eEPCs were used in two
Business Process Improvement projects. The main objective of these case
studies was to find situations in the enterprise models to improve the
reference model. These improvements could be semantic, pragmatic or
syntactic improvements.
These projects were conducted with industry partners in collaboration with
the Queensland University of Technology. In the first half of 2003 these
projects were conducted in the area of IT Service Support specifically
Incident Management. The BPI methodology used was that of the Business
Process Lifecycle (Rosemann 2000), which includes the steps Process
Identification, AS-IS Modelling, AS-IS Analysis, TO-BE Modelling, Process
Implementation, Process Execution and Process Monitoring as discussed in
Chapter 2. Postgraduate IT student teams were given 13 weeks to suggest
improvements to the processes effectively reaching the end of the TO-BE
modelling activity.
The history of these collaborative projects and a description of the general
educational aims of the projects can be found in Rosemann et al. (2000).
Students were given access to both the ITIL books (Service Delivery and
Service Support) and the developed reference model in ARIS (which was
provided as web pages), and were encouraged to examine both sources as a
basis for their domain knowledge. Unfortunately, due to discrepancies in the
versions of ARIS available to the students and the author, the models could
not be delivered for direct editing but only displayed on the web pages (i.e.
the model Ability to Edit was locked).
The pairing of the ITIL books and the models was seen by the research team
as an efficient way of providing a formal well defined and consistent model,
Masters Thesis Chris Taylor
QUT Page 175
and the extended content (e.g. glossaries, extended explanations), provided
by the books, that was demonstrated as so important in Chapter 4.
The students were advised to use the sources as they saw appropriate. That
is, they were made aware of the resources, but their use was not mandatory
nor linked to their marks for the subject.
The aim of the case studies was to examine actual help and Service Desks
to help improve the reference model. In this improvement process three
steps were used which are described in more detail in the data analysis
section using a Best Practice by Composition Design principal as discussed
in Chapter 5.
Selection of Participants
The student teams were selected from high achieving post-graduate students
with a high pass in the pre-requisite subject introducing process modelling
and process improvements.
The industry partners were selected based upon suitability to the reference
model, i.e. they were providing IT services. Both organisations were selected
because of the excellent working relationship between the CITI and the
organisation, as well as previous co-operative research efforts. CITEC was
selected because they provide IT services as an outsourcing partner. Pauls
was selected to provide some insight as to how well the reference model
could be applied to an internal service provider, which is closely related to the
target domain of the reference model (i.e. IT services as an outsourcing
arrangement). Applying the reference model outside its specific domain can
offer interesting comparisons.
Presented in next are brief descriptions of the two case study organisations
and their service support environments.
CITEC
Case study one was conducted with one of the focus group member
organisations, CITEC. CITEC is an external IT services provider with several
Service Desks dispersed geographically and logically. For the project the
Masters Thesis Chris Taylor
QUT Page 176
student team investigated two of the Service Desks. One was seen as the
internal benchmark, while the other was a high volume Service Desk. Due to
resource constraints only the AS-IS processes models were completed.
CITEC’s services include infrastructure management, applications
management, information brokerage, business process outsourcing and
associated professional services. Their Service Desk software was Quantam.
PAULS
Pauls, a member of the Parmalat group, runs IT as an internal service. It also
however fields calls from external retailing customers who use their system to
place orders. Help desk which services approximately 600 calls per week
and is in operation 24x7. The IT environment of Pauls has 650 users over 35
sites via LAN/WAN. Pauls has been using the Quetzal 1.03 Help Desk
package for handling calls and support jobs.
Data Collection
There were two sources of data collected about the use of the reference
models in the projects. The most important was the finished enterprise
specific models of the teams, in both the AS-IS and TO-BE scenario (Pauls
team only). The project teams used ARIS as their modelling tool and the final
models for the different scenarios were the primarily source of data for the
testing phase. The data collect was the final databases from the project
teams, which contained the models in the ARIS toolset.
The second source of data was through surveys of the model users (the
students). The surveys were aimed at examining the students’ opinion of the
reference model in relation to the ITIL books and consisted of several open
ended questions. These surveys were also followed by informal discussions
with the students about their experiences using the reference model. Data
about the use, perceived quality and consistency of both the books and the
models was collected from the project team members via surveys. Questions
focused on drawing comparisons between the original ITIL books (books)
and the ARIS models (models). After preliminary demographic information on
modelling and domain background, the respondents were asked to identify
Masters Thesis Chris Taylor
QUT Page 177
how both the models and books were used in the project (e.g. for process
identification, AS-IS modelling template etc.). The survey went on to ask for
possible improvements in both the books and models in free text questions
and finished with a direct comparison between the models and the books on
a five point scale in relation to given statements (e.g. contained more
ambiguities, limited creativity etc.). The topics of these comparisons were
drawn in part from (Schütte 1998) work identifying the impact of reference
model use and partly on the quality attributes described in Chapter 4. The
survey questions are contained in Appendix C.
Data Analysis of Models
The analysis of the models was conducted in two phases.
The first phase looked at the general similarity of the enterprise specific
models to the original reference model. The second phase examined specific
differences and whether these could be used to improve the reference
model.
First Phase – Degree of Re-use
This phase of the analysis was aimed at determining how much of the
models had been re-used. The first phase of the data analysis was aimed at
determining how much of the models had been re-used for the enterprise
models. The aspects considered in the comparison of the models were the
semantics, objects, naming, branching points, layout, hierarchies and
interactions to other processes.
Two coders examined the reference model and then the enterprise specific
models and rated the degree of agreement on a simple three point scale
(high, medium and low). The reason that these ratings were used is that they
give simple to code and simple to interpret results. Attempts were made to
introduce a more detailed coding scheme, but the analysis was prohibitively
time consuming, often very difficult and did not add any useful data.
A high degree of re-use means that the enterprise model was obviously
based on the reference model, and had a high similarity of semantics and
Masters Thesis Chris Taylor
QUT Page 178
syntactic constructs, as a general rule about 70% of the models concepts
appear to be drawn from the reference model.
A medium level of re-use means that, although it is apparent that certain
parts of the model are the same as the reference model, the reference model
has been reasonably changed. As a simple rule around 30-70% of the
concepts of the model are drawn from the reference model.
A low level of re-use means that, most of the model appears to have been
built from scratch or changed beyond recognition from the reference model.
Less than 30% of the model is based on the reference model.
Specific examples of the models classified in three levels of re-use are
contained in Appendix B.
Second Phase – Improvements to the Reference Model
This phase of the analysis was aimed at improving the reference model.
Following on from the first stage, individual differences were identified. After
this identification each change, or related group of changes, was examined to
determine whether the enterprise specific model contained a generalisable
concept, in other words, whether the change was not a result of the
derivation of the model (e.g. changing “Incident record” to “work request”).
If the differences were not a result of derivation, the changes were examined
to see whether they offered higher syntactic, pragmatic or semantic quality to
the reference model. This decision was made by the coders based on their
knowledge of the specific situations, the domain in general and using input
from other information entities to judge the generalisability of the enterprise
model depictions.
There are 4 ways these differences could impact on the reference model if
they were determined to be relevant.
Firstly, the enterprise specific model contained paths or objects that should
be added to the reference model, for example to cover a situation not already
in the reference model. If this was the case then these new semantic
elements were added to the reference model, which are mostly situations
Masters Thesis Chris Taylor
QUT Page 179
that were not conceived of during the reference model design. This situation
is termed an addition.
Secondly, if the enterprise specific model excluded something that was
present in the reference model and it was determined that it should not be in
the reference model it was deleted. In this case a scenario that was included
in the reference model would never, or only very rarely, happen. This
situation is termed “exclusion”.
Thirdly the semantics may be neither of the two first, where the enterprise
model simply depicts a different way of handling the situation. This is termed
a “modification”. The modifications were seen as semantic improvements to
the reference model based on the enterprise model. Essentially, it meant that
the solution depicted in the enterprise model was better practice than that
depicted in the reference model, as judged by the two coders.
Fourthly, the enterprise specific model had a better pragmatic or syntactic
method of depicting the same semantics. In this case, the content of the
model is essentially the same, although the way of depicting it is different.
For example, a particular modelling object or construct that was confusing or
un-necessary. This situation is termed a “pragmatic change”.
Differences that did not fit one of these categories were ignored as not
relevant, for one of 3 reasons. Either they were a result of the derivation
process that was not deemed to be generalisable, they did not add to the
reference model which is aimed at depicting best practice (for example
manual processes for which there are currently available automated support)
or thirdly they were perceived as modelling errors.
Coding
Due to the subjective nature of both of these phases, two independent
codings were conducted on the data, one by the author and the other by a
coder (Dr Christian Probst) with a high domain and modelling knowledge and
experience. Both coders were closely involved in the projects and understood
the domain, the specifics of the organisations involved and of process
Masters Thesis Chris Taylor
QUT Page 180
modelling. After the codings any differences in the results were discussed
and an agreement was reached.
Data Analysis of Survey
The data collected via the surveys was simply averaged (for the Likert Scale
questions), while the responses to the open questions and the informal
discussions are used to provide some context to the other results.
7.3.2 Case Study Results
The case study results are presented in this section. Firstly, the results of the
analysis of the models are presented, followed by the results from the survey.
Results from Analysis of Models
The first phase analysis is presented per model to highlight the differences in
the degree of re-use in each model. In the second phase, the improvements
phase, results is presented per case study, because it is not important from
which models the improvements were derived.
Case 1 – CITEC
Phase One – Degree of Re-use
The AS-IS CITEC value chain followed the reference model exactly. This
provided a convenient unit of analysis for the comparison, i.e. the models for
each step of the process.
VAC Detect Incident
Record Incident
Assess Incident
Resolve Incident
Close Incident
AS-IS model 1 High Medium Medium Medium Low Medium
AS-IS model 2 High Medium Medium Medium Low Medium
Table 26: Re-Use of models in the CITEC case study
Using the data analysis steps outline in the data analysis section above, the
following changes to the reference model were made.
Phase Two – Improvements to the Reference Model
This section explains the improvements made to the reference model based
on the CITEC enterprise specific model. The type and source model of the
improvements are indicated in brackets.
Masters Thesis Chris Taylor
QUT Page 181
Event Logging (Modification - Both)
The enterprise model showed that not all of the errors recorded in a typical
event log generated by a monitoring system should be regarded as Incidents
or be forwarded to the Service Desk. This would often overwhelm a Service
Desk with many trivial and quite regular and normal events. The reference
model was changed to show that only events that the monitoring systems
had classified as Incidents would be forwarded to the Service Desk.
In one of the models the Line of Service staff are able to manually log an
Incident, without going through the Service Desk. Because this would
improve the response times, as well as the classification accuracy this was
also included in the model. The consequences of this choice are discussed in
Chapter 8 in further detail.
Incident Reporters (Addition – AS-IS 1)
It was noted that not only could a user report an Incident but also one of the
provider staff or a third party could also report the Incident. The effect on the
model was to include this in the description of the organisational unit type
“Reporter”.
Using Event Diagrams (Pragmatic Change - Both)
In the original reference model event diagrams were used. These were
removed in the enterprise model, because they added extra complexity to the
model. The information contained in the event diagrams was simply
transferred to the eEPC. In the reference model they were also removed.
Assessment of the Incident (Pragmatic Change - Both)
The enterprise model showed some of the functions originally in the Assess
Incident in Record Incident. The student team misunderstood the Assess
Incident to mean “find the work around for the Incident”. The Assess Incident
step is very important as it is within this step that the prioritisation and
categorisation take place. To emphasise the importance of the assessment
step and distinguish it from the Develop New Resolution step the VAC, both
steps were explicitly expressed in the value chain.
Masters Thesis Chris Taylor
QUT Page 182
Scope of Incident Management (Addition - Both)
The models contained all types of incoming contact with the Service Desk,
even contact that was not to report an Incident. For example, the model
showed how enquiries were processed.
The effect on the models was to show the grouping of the events in the
Assess Incident step, into the different event types, one of which is Incident.
After this the guidance from ITIL can be used to deal with the Incident.
Expanding the scope of the process effectively reuses the start of the
Incident Management process to also deal with the non-Incident events (e.g.
an enquiry for a phone number, an update on some information received by
the Service Desk). This is also a logical step because the same software and
resources could effectively be used to log events as it could to manage
Incidents.
Automatic Escalation Due to Severity (Addition - Both)
The enterprise models showed that once an Incident had been assessed it
could trigger a hierarchical escalation process if the Incident had been
assessed as having a high level of severity.
Taking this idea further, an integrated and sophisticated Service Desk tool
could track automatically trigger escalations (either hierarchical or to Problem
Management) if a particular set of conditions had been met. Examples of the
conditions that could trigger the hierarchical escalation include the level of
impact of an Incident or the combined level of impact multiple linked
Incidents. Conditions that could trigger the Problem Management process
include the combined impact of the number of Incidents attached to a
Problem record or the impact of a particular Incident (or linked group of
Incidents) that should trigger the Problem Management process to create a
new Problem record, or the combined impact of Incidents linked to a
particular Known Error, that may require the economic efficiency of the
resolution of that Known Error to be re-assessed.
Re-assignment during Restoration (Addition - Both)
Masters Thesis Chris Taylor
QUT Page 183
During the restoration process (not just the diagnosis step) the Incident may
need to be re-assigned, if the resolution was traced to another line of service.
The summary of the types of improvements derived and which model/s they
came from is shown in Table 27.
Additions Exclusions Modifications Pragmatic Changes
AS-IS Model 1 1 0 0 0 AS-IS Model 2 0 0 0 0
Both 3 0 1 2 Table 27: Summary of Improvements derived from CITEC Models
Case 2 – Pauls’ Models
Phase One – Degree of Re-use
The AS-IS CITEC value chain followed the reference model exactly. This
provided a convenient unit of analysis for the comparison, i.e. the models for
each step of the process.
VAC Detect Incident
Record Incident
Assess Incident
Resolve Incident
Close Incident
AS-IS model High Low Low Low Low Low
TO-BE Model High Medium Med Low Low Low
Table 28: Re-Use of models in the Pauls case study
The Pauls models had very low re-use of the reference model, particularly
the AS-IS model. An interesting aspect of this case study was the increased
re-use in the TO-BE model, where 3 out of 5 value chain steps increased in
the level of re-use. Even in the Resolve Incident step, although both models
were rated low, there was more re-use in the TO-BE model.
Phase Two – Improvements to the reference model
Only one difference in the Pauls’ models was incorporated in to the reference
model.
Reporter Interaction (Addition - Both)
One of the models from the Pauls processes specifically showed the client
interaction at the Resolve Incident step. The inclusion of the client
interactions in the reference model was added textually in the attributes of
Masters Thesis Chris Taylor
QUT Page 184
the steps in which major interactions with the user or reporter would normally
take place if needed on an ad hoc basis. That is in the Assess Incident and
Develop New Resolution.
Additions Exclusions Modifications Pragmatic Changes
AS-IS 0 0 0 0 TO-BE 0 0 0 0 Both 1 0 0 0
Table 29: Summary of Improvements derived from Pauls Models
Summary of Survey Results
Five individuals from the student teams who were involved with the projects
responded to the surveys. The CITEC team had only 2 members and the
Pauls team had 5 members. The full survey results can be found in the
Appendix D. Presented here is a summary of the results.
The most interesting finding from the survey was the total number of uses for
the books versus the models. Each respondent was asked for how they used
the models and the books. The models were reportedly used in twice as
many times than the books (18 reports in comparison to 9). However, this
was not reflected in the response to the direct comparison, “Which did you
use most?” in which the response favoured of the books. Reasons for this
could include that the books took longer to understand, or that the models
were used as an introduction to the domain and then the books were
consulted for further detail.
Models were also perceived as easier to read and as providing a better
overview than the books and as allowing better semantic quality of the
models produced during the projects (i.e. TO-BE and AS-IS). The models
allowed better opportunity to check the syntactic quality of the models
produced in the projects by providing an example for comparison. There was
a positive reaction to the models compared with the books when asked which
source contained ideas that could most easily be incorporated into the TO-
BE models.
Sample results are provided in Table 30 (see Appendix D for full details of all
the questions plus standard deviations). These results are those that had an
Masters Thesis Chris Taylor
QUT Page 185
average response of more than greater than 4 or less than 2 (i.e. 1 away
from the mid-point of 3) and hence represent the strong views of the
respondents collectively.
Question Average Rating Actual Value
Books (1) (5) Model
Which gave the best overview? x 4.0
Which one was easier to understand? x 4.2
Which allowed semantic errors to be detected
in the AS-IS models? x 4.0
Which allowed syntactic errors to be detected
in the AS-IS models. x 4.0
Ideas from which material can be easily
incorporated into the TO-BE? x 4.2
Table 30: Sample Results from Survey questions
In the open questions, all respondents confirmed the models added value to
the books, and all indicated that in a future project they would use both the
books and the models.
Overall the results from the survey indicated that the students perceived the
models as at least as useful as the books. Assuming that the books are of a
certain quality, which can be judged by their acceptance in industry, then it
can be interpreted that the models are also of a certain quality for these
projects.
7.3.3 Limitations and Conclusions
The major limitation of the case studies was the subjective nature of the
decisions about what aspects of the enterprise specific models should
influence the reference model. However, using two coders, both with
advanced knowledge of the modelling technique, tool, general and specific
domains, mitigates this limitation.
The other major limitation is the external validity of the findings. The testing
phase could benefit from several more case studies, and more surveys of the
reference model users, especially in different sites which have identified as
being “best practice”, which could challenge the ideas put forward in the
Masters Thesis Chris Taylor
QUT Page 186
reference model, to improve the reference model, and also offer more input
as to the amount of re-use that could be expected using the reference model
for BPM.
Perhaps one of the most severe limitations is the use of students to conduct
the “consultancy engagements”. The students lacked extensive experience in
process modelling and had little to no previous exposure to the domain as
indicated by the responses to the demographic questions about previous
modelling experience and domain experience. This limitation was mitigated
by several factors, including the education provided for the students before
the projects were conducted as well as the coaching by several individuals
experienced in the domain and these types of business process
management projects.
The final limitation noted is the disjoint between the purpose of the model and
the application of the model. Although the model depicts “best practice”, it
was used mainly for capturing the AS-IS states of the processes. It was not
used for a TO-BE model for the first case and the production of a TO-BE
process model for the second case was not the focus of that project (the
focus was instead to determine an appropriate IT solution), and hence didn’t
receive the attention and effort that it could have. Also the reference model
processes rely heavily on an integrated and sophisticated IT support
application suite, the implementation of which was not within the scope of the
projects, hence many of the processes depicted could not be exactly copied
into the TO-BE models.
The above limitations should be considered when interpreting the
conclusions from the case study which is presented next. The conclusions
are structured around the two stages of models produced in the case studies,
the AS-IS and TO-BE models and the impact of the reference model on
these models is discussed.
Masters Thesis Chris Taylor
QUT Page 187
AS-IS Modelling
The results show that the model was re-used more in the AS-IS CITEC case
than the AS-IS Pauls case. This could be an indication of the difference in
sophistication between the IT provision processes of the two organisations.
The particular circumstances of the IT service provider may not necessitate a
high level of measurable and repeatable service, especially in the internal
provider case. This could make some of the processes in the reference
model redundant. For example, there may be no check to see whether calls
are covered by a service level agreement. The “best practice” processes may
be very different to the current processes of IT service management
processes in an organisation, making the reference model only of limited use
for the AS-IS modelling for certain organisations, particularly less mature IT
service providers.
In both cases however the re-use of the model was generally low. This could
be due to one or more of the following reasons.
In Chapter 4, the ability to use the model efficiently was identified as a quality
criterion for reference models. Having a locked reference model means that
the model cannot be edited directly, and hence its application (and therefore
degree of re-use) may be limited. When asked what aspects of the reference
model could be improved, the students asked that it be an “open” (i.e. not
locked) model, that they could edit directly.
Another reason is the current maturity and sophistication of the Incident
Management processes at the organisations.
The third reason is that the models may have been perceived as not useful
for the purposes of the projects. The results from the survey questions,
especially about intention to use in future indicate that the students did
perceive the models as useful.
TO-BE Modelling
The Pauls models showed more re-use in the TO-BE models as compared to
the AS-IS models, and even though some of the individual ratings may not
Masters Thesis Chris Taylor
QUT Page 188
have changed, after discussion both coders agreed that in all steps more of
the reference model had been re-used. Despite this increase in re-use much
of the model was not applied. This could be due to several reasons.
In the TO-BE modelling for simple or internal IT service provision
environment, the reference model may not be widely re-useable because it
does not reflect the desired state of this particular IT service provider. In
other words, the model has been built to deal with a complex IT service
provision domain, with possible multiple clients with different SLA, large
numbers of Incidents, many staff, many products and complex support
requirements, which may not be the case, especially as an internal service
provider.
Another reason is that it may be prohibitively expensive to implement the
reference model, meaning that short or even medium term process goals of
the organisation may differ from the reference model.
7.4 Chapter Summary
This chapter covered the testing of the reference model in process
identification and targeting then separately in business process management
projects. Among the major outcomes were the following points:
• The model had a high degree of re-use at the high-level
• The model had only a medium level of re-use in the BPM projects for
the AS-IS modelling for the external IT service provider
• The model had a low level of re-use in the internal service provider
• The model was perceived by the student teams as being useful
• Reasons for the lower level of reuse (as perceived by the researcher)
o Application to a slightly different domain (i.e. internal IT service
provision in the Pauls case)
o Lack of maturity of the organisational processes
Masters Thesis Chris Taylor
QUT Page 189
Chapter 8: ITSP Reference Model Outputs
8.1 Introduction
The previous chapters outlined the procedure used to design and then test
the reference model. This chapter summarises the outputs of that procedure.
It looks at the outputs of the procedure both in terms of the actual process
models that were developed, and ideas contained therein, and in the lessons
learnt about the procedure of designing reference models using the methods
discussed in the previous chapters.
The reference model itself is contained in an ARIS database.
8.2 Model Outputs
8.2.1 Level 0 Model – Business Process Framework
The entry point to the more detailed levels of the model is the business
process framework, which is Level 0 of the model. The naming of the levels
has been based on the conventions used in SCOR and eTOM, with the
highest levels of abstraction, being denoted as the lowest numbered level.
Therefore the level number gives an indication of the scope of the processes
and the level of detail.
This is the highest abstraction of the model and presents core 19 processes
grouped into 5 areas as shown in Figure 69.
Masters Thesis Chris Taylor
QUT Page 190
Figure 69: Process Groupings of the ITSP Reference Model
These groupings, while actually being the highest level of the model, are so
abstract that they are not given the Level 0 designation, rather the individual
processes as shown in Figure 70 form the Level 0 processes. Note that as
discussed in Chapter 2 and Chapter 6, only the core processes have been
further developed.
SERVICE DEFINITION
INFRASTRUCTURE MANAGEMENT
CUSTOMER RELATIONSHIP MANAGEMENT
SERVICE SUPPORT
SERVICE DELIVERY
ENABLERS
STRATEGY
Masters Thesis Chris Taylor
QUT Page 191
Figure 70: Level 0 Processes of the ITSP Reference Model
The processes identified in the high level model are shown in Figure 70 and
are explained below. As stated earlier, ITIL had a large influence on these
definitions and the structure of the reference model.
Hardware Management
Hardware Management is the process of installing, maintaining, upgrading,
the physical infrastructure, including any networks used to deliver the
application on the client or provider side, but excluding the intermediate
network if any, such as that provided by a third party telecommunications
provider. The management of such a relationship with a telecommunications
provider is covered in the Enablers processes. It may include desktop
support but this is not essential.
Software Management
Software Management is the process of maintaining the IT systems that
support the delivered application, such as the database management
software, operating system or any other software that provides the platform
for, or supports, the delivered application or services.
Service Level Definition
Reporting Definition
Har
dwar
e M
anag
emen
t
Sof
twar
e M
anag
emen
t
Appl
icat
ion
Man
agem
ent
Security & Continuity Management
Billing Definition
Service Levels Monitoring
Configuration Management
Capacity Management
Service Scope Definition
Rel
ease
Man
agem
ent
Ser
vice
Enh
ance
men
t
Billing
Reporting
Site M’ment
Prob
lem
M
’men
t
Inci
dent
M’m
ent
SERVICE DEFINITION
INFRASTRUCTURE SERVICE DELIVERY
Cha
nge
Man
agem
ent
SERVICE SUPPORT
CUSTOMER RELATIONSHIP MANAGEMENT
Masters Thesis Chris Taylor
QUT Page 192
Application Management
Application Management deals with the processes specific to the delivered
application. It involves the maintenance of the application, the technical
changes to the application or functional changes to the system.
Security & Continuity Management
Security & Continuity Management is the process that manages the risk to
the ITSP infrastructure. It involves such aspects as physical and logical
security, failure plans and contingency plans and is aimed at providing and
then executing plans for methods of providing the IT service, if the ITSP
infrastructure is impacted on by an event that reduces its ability to provide
services. Examples of such events include natural disasters, acts of terrorism
etc.
Service Level Monitoring
Service Level Monitoring is the process of collecting and analysing the
metrics needed to demonstrate the service provider’s fulfilment of service
level agreements.
Capacity Management
Capacity management involves all the activities that are aimed at matching
the current and future capabilities of the IT resources (both technical and
human) to cost effectively meet the demands placed upon it. This includes
understanding the current and predicting the future demands and managing
consumer demands. This process plays an important role in investigating the
cost effectiveness of investments in infrastructure, especially in the longer
term.
Configuration Management
Configuration Management deals with ensuring that all the items that make
up the IT infrastructure are described and the details and relationships
between them are maintained in a database. Items should include software,
hardware and any documentation associated with these items. The
definitions of processes, service level agreements, software licenses etc.
should be maintained by Configuration Management.
Masters Thesis Chris Taylor
QUT Page 193
Release Management
Release Management involves all the processes from having an approved
change to the monitoring of the release in the live environment. The release,
or rollout, could be of a new or modified software or hardware component.
The process is focused on protecting the live IT environment and ensuring
that the release is checked and monitored before and after implementation.
Change Management
Change Management involves the changing from one state to another of the
IT infrastructure. It involves the necessary checks and approvals to ensure
that any changes made to the infrastructure, are introduced in a controlled
and documented manner and that these changes bring about their expected
benefits to the service.
Service Enhancement
Enhancement management encompasses the ITIL process of Change
Management as well as other activities. The extra activities include
identifying, investigating and possibly preparing a business case for the
opportunities to supplement existing services, effectively placing a more pro-
active role on the generation of change requests.
Discussions since the process has been renamed and increased in the scope
(from Change management to Enhancement management) indicate that this
is an extremely important area for IT service delivery. It allows the provider to
on-sell or leverage current investments and also the client to benefit from the
experience and expertise of the provider, by identifying new opportunities for
IT in the client enterprise. One of the aims of this process should be to
ensure that the IT service is maximally effective (not only efficient) for the
client and that new capabilities that can be cost justified (either to a particular
client or to the provider organisation itself) are implemented.
The Enhancement Management process deals with ITIL’s Change Requests
as well as Service Requests. This process should investigate the operational
impacts of any proposed change.
Masters Thesis Chris Taylor
QUT Page 194
Client Site Management
An important distinction between the in-house provider and the outsourced
provider is the management of on-site premises. This process captures all
the client site management, such as liaising with clients affected by any
changes, regular updates and meeting to discuss operational and strategic
matters with individual clients. It also includes arranging for incident
resolutions where onsite work is needed, scheduled outages, release
timelines etc. and works closely with Release management.
Billing
The process of billing is extremely important to the IT service provider,
particularly the outsourced provider, but also, in the case of internal cost
accounting or cost recovery, for an internal service provider. It involves the
collection of the data required to produce the invoice, the furnishing the
invoice to the client and dealing with payment and billing enquiries.
Reporting
This process involves gathering all the data and presenting it to the client. As
discussed in the focus groups this process is very important and should be
used as a feedback session from the clients to ensure that the service level
agreements are meeting the business needs of the client. It would interact
with the Service Definition processes and the Enhancement processes.
Problem Management
Problem Management deals with the correction of underlying causes of
Incidents. The problem management process involves preventative
measures as well as analysis of data in order to prioritise and resolve
problems and errors in the IT Infrastructure to ensure smooth future
operation. As well as this pro-active element, Problem management can be
triggered from other processes such as Incident Management.
Incident Management
The main objective of Incident Management is to restore the services as
soon as possible after they fall below agreed service levels. This process
spans the process from event detection through to the closure of an Incident
Masters Thesis Chris Taylor
QUT Page 195
record. It also includes all the processes for initial contact with the Service
Desk (including Incident, service requests, enquiries etc.).
Incident Management also handles all of the incoming contact to the Service
Desk.
8.2.2 Level 1 Model: Incident Management – VAC
The Incident Management process also handles all the incoming contact with
the Service Desk, such as non-Incident reports or enquiries. From the ITSP’s
viewpoint the purpose of the Incident Management process is to restore
services (or provide acceptable alternatives), when the level of IT service
provided by the ITSP does not meet the SLAs.
The secondary purpose of the Incident Management process is to provide a
single point of contact for all incoming interaction with the ITSP’s customers.
From the client’s perspective the major outcome of the Incident Management
process provided by the ITSP is to enable the client’s users to continue work,
and provide all the information associated with the services offered by the
ITSP covered by SLA.
The Incident Management value chain is made up of 6 steps. They outline
the end to end process for managing incidents, from the time when they are
detected to when the incident record is closed.
Figure 71: Level 2 Value Added Chain for Incident Management
In the next section each of the steps of the Incident Management process are
described.
1.1 DetectEvent
1.2 RecordEvent
1.3 AssessIncident
1.6 CloseIncident
1.4 Developnew
Resolution
1.5 RestoreServices
Masters Thesis Chris Taylor
QUT Page 196
8.2.3 Level 2 Models
Detect Incident
The main purpose of this step in the Incident Management process is to
make the ITSP, specifically the Service Desk, who should have access to all
current or potential service outages, aware of the situation where a service
level is not being met, or a situation which might lead to a service level
breach.
Record Incident
The purpose of this step is to capture all the details of the Incident. It is
extremely important to ensure sufficient and accurate information, as this
information is used throughout the rest of the Incident Management
processes, as well as in many other processes, especially Reporting, Billing
and Problem Management.
Assess Incident
This step is one of the most important in the process. The purpose of this
stage is to classify the incident and match it to existing knowledge and
information. Both the classification and matching are extremely important.
Accurate classification is essential to determine the related service level
agreement/s, and to provide a basis for prioritisation and matching to existing
knowledge. The matching is one of the value-add steps for the ITSP.
The richness and completeness of the knowledge of the ITSP and the
effectiveness and efficiency of the re-use of the knowledge are essential to
the ITSP. This knowledge and ways of using it is one of the methods by
which the ITSP is able to provide services to customers at reduced cost
compared to internal service providers. Re-using existing resolutions,
contained in other Incident records, problem records or information supplied
by third-parties, removes the need to develop new resolutions.
Develop New Resolution
The purpose of this step is to create new information regarding how to
restore services. As a service provider builds its knowledge the need for this
Masters Thesis Chris Taylor
QUT Page 197
step should be reduced. This step involves diagnosis and other tasks
requiring highly skilled personnel, meaning that it can be quite costly.
Restore Services
The purpose of this step is to apply the resolution, identified through the
Assess Incident step or recently created in the Develop New Resolution step.
The activities in this step may be extremely varied, as simple as advising the
client to use another printer, to a complicated fix to hardware, hence it is only
modelled very generically.
Close Incident
The Close Incident step is crucial as it ensures the accuracy of the initial
assessment, and that the client perceives that the services have been
restored to agreed service levels.
Assumptions
Although the purpose of the model was to depict best practice, several
instances occur where the definition “best practice” is in a general sense was
extremely difficult. Several assumptions and arbitrary choices were made
during the course of the design of the model.
Depending on the exact situation and factors such as pricing models for
support (e.g. flat fee, or per call basis), organisational culture or the
relationship between the client and provider, different solutions can be
perceived as “best practice”. This section describes the choices made in the
reference model, and the reasons they were selected. In doing so, alternate
methods of implementation are also described, although these were not
modelled in the reference model. They could be as variants. The reason they
were not modelled, however, was due to the time and resource constraints.
The choices made, their potential justification and other options for
implementation are described here.
Underlying Data Integrity
It is assumed in the model that all the underlying data that is used (e.g. the
user contact details, configuration items, service levels etc.) are correct. It is
Masters Thesis Chris Taylor
QUT Page 198
conceivable that confirmation of these details is made part of the Incident
Management process, although assuming that the other processes are
running well (particularly Change and Configuration management) these
checks have been excluded from the model.
Incident Logging
With advances in technology, particularly with the internet, service desk tools
can be opened to clients. The decision to open the logging tool and the
extent to which clients can record and assess their own incidents is a matter
that should be determined in relation to other issues, especially the
consequences of inaccurate recording and assessment.
For example, allowing clients to determine the impact and severity of
incidents may lead to many incidents being prioritised as very important,
negating the usefulness of such a prioritisation structure, or service level
agreements bound to certain priorities.
The advantages of allowing users to log their own incidents are the reduction
of service desk time spent on logging and integration to self service IT
support functions.
Self Service
Some IT support tools, e.g. theGuard! Helpdesk from REALTECH, currently
support a self-service IT support using knowledge bases and search engines,
or context sensitive queries.
The implementation of this option would be attractive for flat-fee pricing
agreement structure arrangement. However, as mentioned in the ITIL
documents allowing or encouraging users to solve their own IT problems can
be counter-productive (CCTA 2000 Section 4.1.7). If users attempt their own
solutions they may complicate the situation. Even if they are successful it
may be less efficient for users, who are not experts and without tools and
training, to be fixing IT failures.
Masters Thesis Chris Taylor
QUT Page 199
Also if the pricing model was a fee per Incident or time based model then the
implementation of a self service solution would probably be a sub-optimal
business decision.
Hence the implementation of self service is not depicted in the model.
Incident Acceptance
Events that are failures in the IT infrastructure but degrade (or could
foreseeably degrade) services covered by the SLA are not classified as
Incidents for the reference model. By only classifying the events that are
covered by an SLA, the provider is focused exclusively on the agreed service
levels. An accurate alignment of the SLA’s to the business needs is essential
for this assumption to work. The alternative is to classify all failures as
Incidents.
Incident Assessment and Closure
Some IT monitoring packages, e.g. theGuard! Application Manager from
REALTECH, can automatically detect and resolve certain Incidents. While
the appropriate utilisation of such systems is obviously best practice, ITIL
offers conflicting advice on how these automated processes should be
integrated with the other processes, particularly Incident Management. The
major discrepancy is in relation to the assessment and closure of these
Incidents and closure of any associated Incident records.
The “Service Desk” chapter of ITIL suggests that Incidents detected in the
infrastructure should be automatically routed to the appropriate person and
the Service Desk only notified of the Incident when an escalation point is
reached (CCTA 2000 Section 4.1.15). However, the Incident Management
chapter implies that all Incidents should be assessed by the Service Desk
(i.e. classified, linked to a service level agreement, prioritised etc.). These
steps would be missing in the case where the Incident is automatically
routed. Even if the assessment steps were performed, either automatically or
by the infrastructure team, the assessment would most likely not be
consistent with those being made by the Service Desk personnel.
Masters Thesis Chris Taylor
QUT Page 200
The decision was made that all Incidents should be classified or at least
checked (using some default values attached to common automatically
initiated Incidents) by human operators of the Service Desk. The closure of
the Incident was also left in the hands of a human operator to ensure that the
Incident has been resolved satisfactorily.
Scope of Incident Management Best practice, as derived from discussions in the focus group, is that all
contact with the Service Desk should be recorded. This allows better
monitoring and control of the Service Desk, and provides data facilitating pro-
active measures to improve the efficiency or effectiveness of the service
provision.
In order to achieve maximum reuse of the models and the resources used to
support the implemented processes, the reference model has extended the
scope of Incident Management to include all contact with the Service Desk,
not just the process from the identification of an event as an Incident.
The models were extended to show the complete end-to-end process of the
detection of a service degradation (or upcoming degradation) to the closure
of the Incident record.
Definition of Incident
ITIL defines an Incident as:
“any event which is not part of the standard operation of a service and which
causes, or may cause, an interruption to, or a reduction in, the quality of that
service.” (CCTA 2000 Section 5.2)
where the “service” here is defined by the SLA’s for the particular customer.
One type of Incident is the “Service Request” which is defined as “every
Incident not being a failure in the IT Infrastructure” and gives an example as
“request for information/advice/documentation” (CCTA 2000 Section 5.2).
Service requests could therefore include requests for services not covered by
a service level agreement.
Masters Thesis Chris Taylor
QUT Page 201
For the model, an event (i.e. any contact with the Service Desk) that is
received at the Service Desk is classified into Incident (as defined by the
definition in ITIL), service request or information request, as separate entities.
This definition is not a trivial matter because some of the major key
performance indicators for IT service provision revolve around Incidents.
Service Level Agreements (SLAs) are often linked to Incident numbers,
recovery times and statuses, and, particularly in an outsourcing arrangement,
these metrics can directly effect monetary payments and continued contracts.
Using an example discussed at the focus groups; defining whether a phone
call to the Service Desk asking for a telephone number is an Incident or not
can affect the relationship with the client. For example, by having a broad
definition of Incident, the number of Incidents in a given time period will be
higher. The number of Incidents is often a key performance indicator to which
financial incentives may be linked. This could be particularly important when
benchmarking within the industry.
8.2.4 Model Output Summary
As discussed in Chapter 2, the desired characteristics of the ITSP BP
reference model are as described in Table 31.
Masters Thesis Chris Taylor
QUT Page 202
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Table 31: Characteristics of an IT Service Provision BP Reference Model
Indicates partially completed
The actual model produced did not provide Intermediate and Task level
models for all the processes (only for Incident Management), but did provide
the entire Complete level model. This could allow further work to progress
using the same framework relatively easily.
The model was based on the ITIL framework. Many of the level 1 processes
were drawn directly from the ITIL processes, however Billing was drawn from
the eTOM framework. ITIL’s Change Management process was expanded
upon using feedback from the focus groups and case studies. The structure
and layout of the level 0 and 1 views were based around the concepts in the
eTOM high level models.
The Intermediate level processes (Level 2 - VAC) were derived from the ITIL
Incident Management chapter along with feedback from the focus groups.
Masters Thesis Chris Taylor
QUT Page 203
Another important influence was an enterprise specific model supplied to the
researchers by one of the focus group members.
The Task level models, level 3, were again derived mainly from ITIL however,
they make several assumptions as outlined above. Many of the items have a
description which is derived or extended from the ITIL documents. Other
influences at this level include the enterprise specific model provided by the
focus group member, and software documentation and demonstration,
particularly the SAP Solution Manager, and theGuard! Helpdesk application.
The model provided some guidance in the runtime (as descriptions of the
objects in the model) and the Model Explanation has been explained in the
modelling conventions in this thesis. Several of the ITIL books, particularly
“Planning to Implement Service Management” (CCTA 2002), could be used
to provide the extended content in the Implementation and Run-Time areas.
8.3 Methodological Outputs
This section examines the perceived efficiency and effectiveness of the use
of the reference model design procedure. It examines what went well, what
didn’t and most importantly the lessons learned about the design of a
business process reference model.
8.3.1 Use of Focus Groups
This subsection examines the use of focus groups for reference model
design, looking at what worked well, and provides suggestions for future work
in using focus groups for reference model design.
High Level Modelling
The focus groups worked well for the higher level models. This could be due
to the higher level of participants (middle management) who had a good
overview of the operational aspects of their organisations. Due to its highly
generic level, there is little room to suggest best practice and the
development of the model was relatively straight forward.
Masters Thesis Chris Taylor
QUT Page 204
Best Practice Thinking
Difficulties were encountered at the lower levels. At this level of detail
participants found it difficult to, in the words of Frank, “transcend their
subjective preferences and attitudes” (Frank 1999 p. 697). The purpose of
producing best practice, i.e. no constraints thinking, may not have been clear
to the participants, or the participants were not able to take this ‘green-field’
approach. A sense of defensiveness may have been felt by the participants.
This may have occurred because they were in front of their industry peers
and ultimately their competitors in some instances. The opinion of the
research team is that the participants were not consciously trying to defend
their positions; however this seemed to influence their thinking. Suggestions
for improvement would be to spend more time setting the directions and
perspective of thinking.
A major limitation on the ideas that generated from the focus group were the
constraints on the IT systems supporting the Incident management in
particular.
Discussing Detail
Problems were also encountered when modelling the detailed level
processes. This could be due to the lack of focus by middle management on
the detail of the processes. Participants at a lower level, for example Service
Desk staff or supervisors, could have been more helpful in the area of
Incident management because their sole responsibility lies in this area, as
opposed to the management level whose focus is spread across the
operations of the organisation.
Suggestions for improvements include inviting lower level employees who
would have an intimate knowledge of the problems and possible solutions
and improvements in the process that could contribute to the development of
best practice at this level. Because they may not be responsible for the
design of the process, only its execution, they may also view the process
more objectively. Another advantage is that those who execute the process
may be more able to ignore the implementation costs of best practice than
middle management. Ignoring the implementation costs and focusing on the
Masters Thesis Chris Taylor
QUT Page 205
end state of best practice may make the focus group more creative and
appreciative of new ideas.
Attendance
Focus group attendance was a problem. Sometimes, of the 6 or 7 confirmed
attendants, as few as 4 actually attended. This was due largely to the time
constraints on the participants. Thought was given to individual sessions but
that negates the advantages of the focus group which include the discussion
between the participants which supposedly provides value to a reference
model.
One suggestion for improving the attendance is to invite more people than
needed (e.g. if aiming for 10 participants aim for 14 confirmations).
Domain Vocabulary
One of the more difficult aspects of managing the focus groups was the use
of domain specific vocabulary. Kitzinger when discussing focus groups
mentions that the research method is useful because participants can
discuss the topic “in their own vocabulary” (Kitzinger 1995 p300). Often the
participants used different vocabulary when discussing the topics.
In future a session could be run specifically on the vocabulary to be used for
the focus group, or information and definitions could be sent to the focus
group members before the focus group. Alternatively the focus group
moderator could spend time with the audience members individually before
the focus group and during the focus group act as an interpreter between the
organisations.
The other option is to develop a glossary of terms when problems are
encountered, although this may distract the focus group.
8.3.2 Bottom-up versus Top-down
The modelling was conducted in a top down approach. The Level 0 is more
generic so it is difficult to engender any ‘best-practice’ ideas into the model.
Also at this level it was relatively easy to reach agreement on the model.
When the level 2 models were constructed however, they sometimes didn’t fit
Masters Thesis Chris Taylor
QUT Page 206
neatly into the higher levels. When this was the case, changes were made to
the level 1 model.
It is thought a combination of the top down approach and bottom up, with
validation and modification using the lower level of detail would be the best
approach for this purpose.
8.3.3 Modelling of Process Alternatives
During the model design, the determination of the “best practice” was made
difficult due to the varying circumstances with which organisations are faced
and the environment in which they operated (specific examples of which
were provided in Chapter 7). Because the scope of the modelling was
reduced, this research didn’t question certain aspects of IT service provision.
For instance, this research didn’t look at whether it was best practice to
charge per instance, per hour or flat fee for service support. At first, the
decision about billing models does not appear important in the scope of the
designed low level model (i.e. Incident Management) until one considers the
“best practice” examples described in Chapter 7, for example the
implementation of self service.
Self-service should cut the number of calls to the Service Desk, reducing the
service provider’s costs. This can increase profits in a flat fee service
provision environment but could lower them in a per usage or per hour
charging model. Essentially, the decision about the validity of self service as
“best practice” IT support is dependent on the choices made about the best
practice for billing models, at least in part.
Similar examples can be found throughout the models, where choices
depend on other choices made within the model, or even external to the
model, such as market demand or regulatory requirements. Some way of
depicting or at least capturing these dependencies would be advantageous.
A way of linking, and depicting this linking, of process models to different
choices made either in other parts of the model or external to the model is
needed. This complements the research underway by Rosemann (2002) on
the customisation of reference models.
Masters Thesis Chris Taylor
QUT Page 207
Chapter 9: Summary and Outlook
9.1 Thesis Summary
This thesis has concentrated on two topics. Firstly, and most prominently,
was the area of reference models. The classification scheme, presented
again in Table 32, answered the first research question:
RQ2. What types of reference models exist?
Characteristic Type Characteristics
View Data/ Information
Process/ Behaviour Function Organisational
Language Formality Natural Language Meta-Model Ontological Theory
Level of Detail Complete Intermediate Task
Focus Business Technical Application
State Common Practice Best Practice
Functional Area Function Specific Enterprise Inter-organisational
Economic Activity (Industry) Org. Specific EA Specific General
Tool Support Producer supplied Third party supplied Public Domain
Extended Content Implementation Run-Time Model Explanation
Readiness for use Single depiction Contains variants Abstract
Distribution Public Domain
Proprietary not available
Available for purchase
Available through
membership
Ability to Edit Locked Open
Currency Living Discontinued Static
Table 32: Reference Model Classification Characteristics
This classification scheme is based on previous classification schemes, from
enterprise models and reference models in particular. The scheme integrated
several of the previous schemes and extended them after reviewing several
reference models. It also provided explanations of each of the characteristics
and characteristic types.
Masters Thesis Chris Taylor
QUT Page 208
Several models, related to the second topic IT service provision, were
discussed and classified using the scheme. The classification scheme
provided an overview of the types of reference models and also language
with which to discuss them. This language was then used to describe how
several different types of reference models can be applied in enterprise
modelling. After a general description, the application of reference models in
a particular context was described.
The next contribution was the answer to the third research question.
RQ3. How can business process reference models be used for process management?
The Business Process Management lifecycle was introduced to provide
specific examples of how and where business process reference models can
be applied and is summarised in Figure 72.
Figure 72: Business Process Lifecycle and related use of Reference Model
Select Reference Model
Process Identification High-Level Template
Process Targeting Scope definition/ suggestions
AS-IS Modelling AS-IS Template
Analysis Process Benchmark
TO-BE Modelling TO-BE Template
Process Implementation
Implementation information
Process Execution
Monitoring/Control Run Time
Information/Performance Benchmark
Masters Thesis Chris Taylor
QUT Page 209
Chapter 3 introduced and justified the research methodologies being used in
the research, namely case studies and focus groups.
After identifying the gap in the literature regarding information on what
classifies quality in reference models, two case studies were performed,
aiming at defining, from a user perspective, the quality aspects of business
process reference models. The conclusions from that research not previously
mentioned in the literature or contradicted the literature are presented in
Figure 73.
Figure 73: Quality Conclusions that add to or differ from literature
This answered the second question;
RQ4. What characterises the quality of a business process reference models?
Chapter 5 defined several design philosophies that could be employed to
design reference models with specific characteristics. The philosophies are:
Blue Sky Design, Design by Abstraction, Design by Choice, Baseline Design,
Common Practice Design, Best Practice by Composition and Explicit
Alternatives Design. The advantages and disadvantages of each of these
philosophies were briefly discussed and the philosophies were integrated into
the reference model design procedural model. The procedural model is
shown again in Figure 74.
This chapter answered the question;
Syntactic Quality • Glossary • Simple
Semantic Quality • Scope and
limitations • Validation of
models • “Over-
completeness”
Pragmatic Quality • Application
documentation • Vendor training • Feedback
mechanisms
Economic Efficiency of Use
Masters Thesis Chris Taylor
QUT Page 210
RQ5. How can a reference model be produced?
The design procedural model was then used to design a partial reference
model for IT service provision, particularly in Incident Management.
Masters Thesis Chris Taylor
QUT Page 211
Figure 74: Proposed BP Reference Model Design Procedural Model
DesignHigh-Level
ValidateHigh-Level
TestHigh-Level
DesignLow-Level
ValidateLow-Level
SelectLow-Level
TestLow-Level
No changesmade
Low Level OK,RM finished
DefineReference
Model
Changesmade
Low Levelchanged
Low Level OK,RM notfinished
Reference Model to be
designed
Masters Thesis Chris Taylor
QUT Page 212
Chapter 6 outlined the steps used to design the reference model, using the
procedural model as a basis, the high-level view of which is presented in
Figure 75.
Figure 75: Level 0 Processes of the ITSP Reference Model
Chapter 7 outlined the testing of the reference model, both at the high level,
using Huxley’s case studies, and then using the lower level reference model
in 2 business process management projects. The testing helped identify
areas that needed work in the model and also examined how much of the
model was re-used. Levels of re-use were generally low, although feedback
from the student participants indicated that they perceived the reference
models were useful. Several conclusions were put forward to explain this
apparent contradiction. One of the projects was with an internal service
provider, while the model was built for an outsourcing agreement, indicating
that the outsourcing processes may not suit an internal service provider. The
models provided to the consultancy team were “locked”. The final conclusion
on this issue is that the “best practice” depiction did not acknowledge the cost
of implementation of such practices, this may have reduced the desire to
implement such solutions, at least in the short to medium term, as depicted
by the TO-BE models.
Service Levels Definition
Reporting Definition
Har
dwar
e M
anag
emen
t
Sof
twar
e M
anag
emen
t
Appl
icat
ion
Man
agem
ent
Security & Continuity Management
Billing Definition
Service Levels Monitoring
Configuration Management
Capacity Management
Service Scope Definition
Rel
ease
Man
agem
ent
Ser
vice
Enh
ance
men
t
Cha
nge
Man
agem
ent
Billing
Reporting
Site M’ment
Prob
lem
M
’men
t
Inci
dent
M’m
ent
SERVICE DEFINITION
INFRASTRUCTURE SERVICE DELIVERY CUSTOMER RELATIONSHIP
SERVICE SUPPORT
Masters Thesis Chris Taylor
QUT Page 213
The design and testing of the reference model answered the question:
RQ1: What is an appropriate business process reference model for IT service provision?
The final stages of the thesis presented the research team’s view on the
lessons learnt from the process. The output from the procedure of designing
a reference model was discussed, as well as the procedure itself.
Fourteen high level processes were defined, and one of these processes was
described with lower level models, i.e. Incident Management. Several of the
process assumptions were outlined. The assumptions themselves as well as
their implications were discussed, in order to provide the reference model
user some awareness that in certain circumstances “best practice” was
highly dependant on other factors, both internal and external to the IT service
provider.
With the benefit of hindsight the use of a “Contains Variants” model, would
have been beneficial to capture the alternate perceptions of “best practice”
and the reasoning behind each perception. The acknowledgement that
different circumstances resulted in different “best practices” may have also
helped in the focus group setting, allowing differing points of view to be
acknowledged and captured for further analysis.
As well as the actual content of the models, lessons were learned about the
procedure of reference model design itself, particularly in relation to the use
of focus groups.
Suggestions that focus group moderators define the language for the focus
group and spend more time setting the perspective of the participants, i.e. to
rely on their experience but not be constrained by it, tighter moderation of the
focus groups, and the use of lower level participants were all put forward as
ways of improving the method.
9.2 Future Work
There are several avenues for future follow on work from this thesis.
Masters Thesis Chris Taylor
QUT Page 214
From a general sense regarding the methodology of reference models, the
design philosophies could be further detailed, integrating methods of
comparison, combination and rating for information entities, and developing
modelling tools to support the reference model design procedure.
The reference model design procedure could be used to design reference
models in other domains.
Also methods could be found to combine common types of reference models,
for example process models with performance measurement models, or
extend the process view into other views including organisational and data.
Some work that is currently taking place using existing enterprise systems to
“re-document” models from actual activity e.g. IDS “Re-documentation Scout”
or the paper by Weijters and va der Aalst (2002), could be combined with
reference models, to automatically identify gaps between an organisation’s
actual processes and best or common practice processes. This could
provide industry benchmarking or standardisation data. If the reference
models had been integrated with a Maturity Model, then the maturity level of
an organisation’s processes could be automatically, or at least semi-
automatically, calculated. This could then form the basis of quality
accreditation (e.g. ISO9001), or assertion of compliance with other standards
(e.g. BS15000 the recent British standard for IT service management).
A major step forward for reference models would be the design of a
modelling language and tool support for the explicit alternatives type of
reference models which is the topic of research currently underway. Not only
should these explicit alternatives be able to be depicted and analysed in a
model, along with their inter-dependencies, but decision support tools could
be integrated to help those deriving the enterprise specific models chose
which alternatives were appropriate for the situation. Information including
case studies, success stories, associated vendor support tools or consulting,
cost benefit analysis templates, critical success factors could all be linked to
specific business questions, the answers to which could automatically
generate the enterprise specific model based on the reference model. In
Masters Thesis Chris Taylor
QUT Page 215
effect this would create an upper-CASE tool, not for software, but for process
models, a CAME (Computer Aided Modelling Engineering).
For example, a question could be “would you like to implement a self service
tool?”. Provided with the question would be available industry solutions,
costs, benefits, case studies etc. By answering “yes” the model would
automatically add the self service processes to the enterprise specific model,
answer “no” would remove them. Interactions could also be mapped, the
information presented in the above could be different depending on the
answer to the question “would you like to bill on a per use, flat rate or per
hour basis?”. The answer to which would not only change the model but
influence the question and information of the self service question. Such a
model could also be used after implementation, as a scalable solution to
changes in business needs, by identifying how business decisions will/should
impact business processes (e.g. by changing the pricing model).
In the shorter term, the design of the ITSP reference model that has been
produced can be continued, adding the lower levels of detail to the other
processes, with further testing. Data, organisational and other views could be
added as well as an implementation guide or procedural model. To help with
the implementation, which for ITIL is often done a process at a time, an
“process interaction model” may be useful. This shows all the process
interactions that a single process has with the other processes. The data
dependencies could also be added. An example of such an interaction model
is shown in Figure 76.
For example in Figure 76, the main process being explored is centred. The
incoming and outgoing events, and the interfaces to the other processes, are
shown as connecting to this central process. This allows an easy reference to
highlight the interactions between the central process and other processes.
Masters Thesis Chris Taylor
QUT Page 216
Figure 76: Example of possible Interaction Model
On a commercial level, reference models could be used for the design,
selection or evaluation of a IT service management software. This interest
was sparked by comments made during the focus group that a major
hindrance to the streamlining of IT service provision was the lack of
integrated IT support. A best practice solution will integrate IT support for the
process, in much the same way that the original BPR thinking was closely
linked to the introduction of integrated computing solutions in other areas of
business processes.
Within the model a distinction is made between manual (or semi-manual)
steps and automated steps, defining that best practice is as automated as
possible. Another model has been added to show how the business
processes could be used to derive data models, that would aid in software
development for the domain.
The body of the thesis concludes with these outlooks on possible future work.
Next are the references, followed by the appendices.
HardwareManagement
SoftwareManagement
ApplicationManagement
System erroralert
User detectsevent notin support
enabled application
User detectsevent in support
enabled application
MonitoringApplication
detects Incident
Event is ServiceRequest
EnhancementProblemManagement
Error causing Incidentcannot be identified
1. IncidentManagement
Enhancement ProblemManagement
Service RequestCompleted
Problemresolved
Incident Closed
Diagnosis found
needs updating
Masters Thesis Chris Taylor
QUT Page 217
References Anonymous 1993. CIM-OSA: Open System Architecture for CIM, Springer, Berlin.
Anonymous 1998, AS/NZS 2777.1:1998, Information processing systems - Open Systems Interconnection - Basic reference model - The basic model Australian Standards.
Anonymous 2001, Technical Reference Model User Guide US Department of Defense. Location: http://trm.disa.mil/userguide_11apr01.pdf Assessed: 18/05/2002
Anonymous 2002, ASC X12 Reference Model for XML Design ANSI ASC X12C Communications and Controls Subcommittee, Data Interchange Standards Association, Inc. http:// www.x12.org/x12org/comments/ X12Reference_Model_For_XML_Design.pdf Assessed 19/07/2002
Anonymous 2003. AS/NZS 14143.4:2003 ISO/IEC TR 14143-4:2002 Australian/New Zealand Standard Information technology—Software measurement—Functional size measurement Part 4: Reference model.
Anonymous Year Unknown, SAP Help "R/3 Reference Model", SAP. Location: http://help.sap.com/sapdocu/core/470/helpdata/EN/35/28ea32e8aa5570e10000009b38f983/content.htm, Assessed: 14/05/2002.
Barry, P. 2002, 'How One.Tel fiddled the books', The Age, 9 April, pp.
Becker, J., M. Rosemann and C. v. Uthmann 2000, Guidelines of Business Process Modeling. Location: http://www.leonardo.com.au/library/Paper%20Becker%20Rosemann%20Uthmann.pdf, Assessed: 14/05/2002.
Bell, D. 1973. The Coming of Post-Industrial Society, Basic Books, New York.
Benbasat, I., D. K. Goldstein and M. Mead 1987, 'The Case Research Strategy in Studies of Information Systems', MIS Quarterly, vol. 11, no. 3, pp 18.
Bennett, C. and G. Timbrell 2000, 'Application Service Providers: Will They Succeed?' Information Systems Frontiers, vol. 2, no. 2, pp 195-211.
Benson, K. 2002, Australia IS Outsourcing Services - IS Outsourcing Competitive analysis IDC, 63.
Bernus 1998, GERAM : Generalised Enterprise Reference Architecture and Methodology IFIP-IFAC Task Force.
Bernus, P., K. Mertins and G. Schmidt, Eds. 1998. Handbook on Architectures of Information Systems. Berlin, Springer.
Biemans, F. 1990. Manufacturing Planning and Control: A reference model, Elsevier Science Publishers, Amsterdam.
Boyle, R. D. 2002, 'Doing business with an application service provider', Strategic Finance, vol. 83, no. 9, pp 24.
CCTA 2000. Service Support CD, The Stationary Office, London.
CCTA 2002. Planning To Implement Service Management, The Stationary Office, London.
Chin, D., K. Takea and I. Miyamoto 1989, 'Using Natural Language and Stereotypical Knowledge for Acquisition of Software Models', in K, Mayo (eds.), Proceedings of
Masters Thesis Chris Taylor
QUT Page 218
IEEE International Workshop on Tools for Artificial Intelligence. Architectures, Languages and Algorithms., pp 290-295.
Curran, T., G. Keller and A. Ladd 1998. SAP R/3 Business Blueprint: Understanding the business process reference model, Prentice Hall, New Jersey.
Curran, T. and A. Ladd 2000. SAP R/3 Business Blueprint: Understanding enterprise supply change management, Prentice Hall, New Jersey.
Curtis, B., M. Kellner and J. Over 1992, 'Process Modeling', Communications of the ACM, vol. 35, no. 9, pp 75-90.
Davenport, T. 1993. Process Innovation: Reengineering work through information technology, Harvard Business School Press.
Davenport, T. and L. Prusak 1997. Information Ecology: Mastering the Information and Knowledge Environment, Oxford University Press, Boston.
Davis, R. 2001. Business Process Modelling with ARIS: A practical guide, Springer, London.
Drucker, P. F. 1990. The new realities : in government and politics - in economy and business - in society - and in world view, Mandarin Paperbacks, London.
Dubie, D. 2002, 'Reaping the rewards of best practices', Network World, vol. 19, no. 39, pp 10.
Duffy, T. 2001, 'Efficiency tool kit', Network World, vol. 18, no. 45, pp 56.
Dunn, P. 2001, 'ASPs: The lure of task outsourcing', Hospitals and Health Networks; Chicago, vol. 75, no. 8, pp 38-40.
Emam, K., J. Drouin and W. Melo, Eds. 1998. SPICE: The Theory and Practice of Software Process Improvement and Capability Determination. Washington, IEEE computer Society.
Fabbrini, F., M. Fusani, S. Gnesi and G. Lami 2000, 'Quality Evaluation of Software Requirements Specifications', in (eds.), Proceedings of 13th International Software Quality Week Conference, Software Research Institute, pp
Falkenberg, E. D., W. Hesse, P. Lindgreen, B. E. Nilsson, J. L. H. Oei, C. Rolland, R. K. Stamper, F. J. M. V. Assche, A. A. Verrijn-Stuart and K. Voss 1998, Framework of Information Systems Concepts IFIP.
Fern, E. F. 2001. Advanced focus Group research, SAGE Publications, Thousand Oaks, California.
Fettke, P. 2003. Personal Communication Reference model classification. Email. 2/04 To: C. Taylor.
Fettke, P. and P. Loos 2003, 'Classification of reference models: a methodology and its application', Information Systems and e-Business Management, vol. 1, no. 1, pp 35-53.
Fowler, M. 1997. Analysis Patterns - Conventions of Thought, Addison Wesley Longmann, Massachusetts.
Frank, U. 1999, 'Conceptual modelling as the core of the information systems discipline - perspectives and epistemological challenges', in Goodhue (eds.), Proceedings of Fifth Americas Conference on Information Systems, Milwaukee, pp
Masters Thesis Chris Taylor
QUT Page 219
Gable, G. 1991. Consultant Engagement Success Factors: A case study and survey of factors affecting client Involvement in, and satisfaction with, consultant engagement in computer system selection projects, carried out for the Small Enterprises Computerisation Programme in Singapore, University of Bradford.
Gable, G., G. 1994, 'Integrating Case Study and Survey research methods: an example in Information Systems', European Foundation of Information Systems, vol. 3, no. 2, pp 112-126.
Garschhammer, M., H. Hauck, B. Hegering, I. Kempter, I. Radisic, H. Rolle, H. Schmidt, G. Hegering, M. Langer and M. Nerb 2001, 'Towards generic Service Management: Concepts A Service Model Based Approach', in (eds.), Procedings of Integrated Management, Washington, pp 301-4
Green, P. and M. Rosemann 2000, 'Integrated Process Modelling; An Ontological Evaluation', Information Systems, vol. 25, no. 2, pp 73-87.
Guldentops, E., J. Lainhart, E. Schuermans, J. Beveridge, M. Donahue, G. Hardy, R. Saull and M. Stanley 2000, Corbit 3rd Edition Control Objective Corbit Steering Committee and the IT Governance Institute.
Gulla, J. A. and T. Brasethvik 2000, 'On the Challenges of Business Modelling in Large-Scale Reengineering Projects', in K Smith (eds.), Proceedings of 4th International Conference on Requirements Engineering, pp 17-26.
Gupta, U. and G. A 1992, 'Outsourcing the IS Function: Is it Necessary for your organisation?' Information Systems Management, vol. 9, no. 3, pp 44-50.
Hammer, M. 1990, 'Reengineering Work: Don't Automate, Obliterate', Harvard Business Review, vol. 68, no. 4, pp 104.
Hammer, M. and C. Champy 1993. Reengineering the Corporation: A Manifesto for Business Revolution, Harper Business, New York.
Hars, A. 1993, Reference data models - Structure and application of libraries for semantic data models. Location: http://www-rcf.usc.edu/~hars/thesis_e.htm, Assessed: 13/05/2003.
Hay, D. C. 1996. Data Model Patterns - Conventions of Thought, Doreset House Publishing, New York.
Helbig, R. 1999, 'Process Management in the Agricultural Commodity Trade and Consequences on Vertical and Horizontal Co-operation', in K Hallow (eds.), Proceedings of 1999 World Food and Agribusiness Congress: Agribusiness Forum, Florence, Italy, International Food and Agribusiness Management Association, pp
Hollingsworth, D. 1995, The Workflow Reference Model, Workflow Management Coalition. Location: http://www.wfmc.org/standards/docs/tc003v11.pdf, Assessed: 30/4/2002.
Huxley, C. 2003. Implementing an Improved Method to Identify Critical Processes. Brisbane, Queensland University of Technology.
Ivancevich, J., P. Lorenzi, S. Skinner and P. Crosby 1997. Management: Quality and Competitiveness, Irwin/ McGram-Hill, Boston.
Kang, I.-S., J.-H. J. Bae and J.-H. Lee; 2002, 'Database semantics representation for natural language access', in (eds.), Proceedings of First International Symposium on Cyber Worlds, pp 127-133.
Masters Thesis Chris Taylor
QUT Page 220
Kaplan, R. S. and D. P. Norton 1992, 'The Balanced Scorecard: Measures that Drive Performance', Harvard Business Review, vol. 70, no. 1, pp p9.
Kaplic, B. and P. Bernus 2001, 'Business Process Modelling in Industry: The powerful tool in enterprise management', Computers in Industry, vol. 47, no. 3, pp 299-318.
Kara, D. 2002, 'Lining up behind ITIL', Software Magazine, vol. 22, no. 1, pp 68.
Kitzinger, J. 1995, 'Qualitative Research: Introducing focus groups', British Medical Journal, vol., no. 311, pp 299-302.
Krogstie, J., O. I. Lindland and G. Sindre 1995, 'Defining Quality Aspects for Conceptual Models', in Unknown (eds.), Procedings of IFIP International Working Conference on Information Systems, Chapman and Hall, pp 216-231.
Lacity, M. and L. Willcocks 2000, Inside Information Technology Outsourcing: A state-of-the-Art Report Templeton Research.
Lavery, R. 2001, 'Does application hosting make sense for you?' Strategic Finance, vol. 83, no. 3, pp 44.
Lee, A. S. 1989, 'A Scientific Methodology for MIS Case Studies', MIS Quarterly, vol. 13, no. 1, pp 33-52.
Lindland, O. I., G. Sindre and A. Solvberg 1994, 'Understanding quality in conceptual modeling', IEEE Software, vol. 11, no. 2, pp 42-49.
Loos, P., O. Keitzel and A.-W. Scheer 1996, 'Adaptable Information Systems by Generic Structures', in G. Ernst (eds.), Proceedings of Sixth Workshop on Information Technologies and Systems - WITS'96, Cleveland, Ohio, pp 258-267.
Lowe, D. and R. Webby 1999, 'A Reference Model, and Modelling Guidelines for Hypermedia Development Processes', The New Review of Hypermedia and Multimedia, vol. 5, no., pp 133-150.
Malone, T., K. Crowston, J. Lee and B. Pentland 1999, 'Tools for inventing organizations: Toward a handbook or organizational processes', Management Science, vol. 43, no. 3, pp 425-443.
McGowan, C. L. and S. A. Bohner 1993, 'Model based process assessments', in Unknown (eds.), Proceedings of 15th International Conference on Software Engineering, pp 202-211.
Mertins, K. and P. Bernus 1998, 'Reference Models', in G. Schmidt (ed.), Handbook on Architectures of Information Systems. Springer-Verlag, Berlin, pp 615-617.
Merton R, F. M., Kendall P. 1956. The focused interview: a report of the bureau of applied social research., Columbia University, New York.
Miles, M., B., and A. Huberman, M. 1984. Qualitative data analysis: a source book of new methods, Sage publications.
Misic, V. B. and J. L. Zhao 2000, 'Evaluating the quality of reference models', in Unknown (eds.), Proceedings of 19th International Conference on Conceptual Modeling, Salt Lake City, Utah pp 484-498
Morgan, D. L. 1988. Focus Groups as Qualitative Research, SAGE Publications, Newbury Park, California.
Masters Thesis Chris Taylor
QUT Page 221
Morin, R. M. 1999, 'British invasion: Will ITIL take hold in the U.S.?' Service News, vol. 19, no. 1, pp 21.
Myers, M. D. 1997, 'Qualitative Research in Information Systems', MIS Quarterly, vol. 21, no. 2, pp 241-242.
Niessink, F., V. Clerc and H. van Vliet 2002, The IT Service Capacity Maturity Model Software Engineering Research Centre.
O'Neill, J., B. B. Small and J. Strachan 1999, 'The Use of Focus groups within a Participatory Action Research Environment', in L. A. Suzuki (ed.), Using Qualitative Methods in Psychology. SAGE Publications, Thousand Oaks, California, pp 237.
Perdue, J. M. 2000, 'ASPs sink fangs into IT costs', Oil & Gas Investor, vol., no., pp 19.
Porter, M. 1980. Competitive strategy : techniques for analyzing industries and competitors, Free Press, New York.
Porter, M. E. 1985. Competitive Advantage, The Free Press, New York.
René Gaches Year Unknown, CimOsa Framework. Location: http://www.rgcp.com/cimosa.htm, Assessed: 5/5/2003.
Reyneri, C. 1999, 'Operational building blocks for business process modelling', Computers in Industry, vol. 40, no. 2-3, pp 115-123.
Rosemann, M. 1998, 'Managing the Complexity of Multi-perspective Information Models Using the Guidelines of Modelling', in Unknown (eds.), Proceedings of 3rd Australian Conference on Requirements Engineering, Deakin Australia, pp
Rosemann, M. 2000, 'Using Reference Models within the Enterprise Resource Planning Lifecycle.' Australian Accounting Review, vol. 10, no. 3, pp 19-30.
Rosemann, M. 2002, Process-oriented Administration of Enterprise Systems, Queensland University of Technology. Location: http://www.citi.qut.edu.au/research/ism/projects/admin_enterprise.jsp, Assessed: 05/02/2003.
Rosemann, M., I. Davies and P. Green 2003, 'The very model of modern BPM', Information Age Feb/March: pp.
Rosemann, M. and P. Green 2000, 'Integrating Multi-Perspectives into Ontologies.' in W. J. Orlikowski (eds.), Proceedings of 21st International Conference on Information Systems, Brisbane, pp
Rosemann, M., D. Sedera and W. Sedera 2000, 'Industry-oriented Education in Enterprise Systems', in G. Gable (eds.), Proceedings of 11th Australasian Conference on Information Systems, Brisbane, Vitale, pp CD-ROM.
Ross, J. W., M. R. Vitale and C. M. Beath 1999, 'The Untapped Potential of IT Chargeback', Management Information Systems Quarterly, vol. 23, no. 2, pp.
Ryoo, J., J. F. Stach and E. K. Park 1999, 'Extension and partitioning of use cases in support of formal object modeling', in Unknown (eds.), Proceedings of Application-Specific Systems and Software Engineering and Technology, pp 238-243.
Saulnier, C. 2000a, 'Groups as data collection method and data analysis technique: Multiple perspectives on urban social work education', Small Group Research, vol. 31, no. 5, pp 607-627.
Masters Thesis Chris Taylor
QUT Page 222
Saulnier, C. F. 2000b, 'Groups as data collection method and data analysis technique: Multiple perspectives on urban social work education', Small Group Research, vol. 31, no. 5, pp 607-627.
Scheer, A. 1994. Business Process Engineering: Reference Models for Industrial Enterprises, Springer-Verlag, New York.
Scheer, A. 1998, 'ARIS', in G. Schmidt (ed.), Handbook on Architectures of Information Systems. Springer-Verlag, Berlin, pp 541-565.
Scheer, A. 1999. Business Process Frameworks, Springer-Verlag, Berlin.
Schütte, R. 1998. Grundsätze ordnungsmäßiger Referenzmodellierung, Gabler, Wiesbaden.
Seymour, L. A. and M. Edwards 2001, Customer Value Powers Opportunity for Network and Hosting Service Providers IDC White Paper Sponsored by IBM.
Silverston, L., W. H. Inmon and K. Graziano 1997. The Data Model Resource Book - A library of logical data models and data warehouse designs, John Wiley & Sons Inc., New York.
Sofaer, S., B. Kreling, E. Kenney, E. K. Swift and T. Dewart 2001, 'Family Members and Friends Who Help Beneficiaries Make Health Decisions', Health Care Financing Review; Washington, vol. 23, no. 1, pp 16.
Sriram, V. Year Unknown, Challenges in the Asian market. Location: http://www.infosys.com/investor/analystmeet2002/ppt/Challenges-in-the-asian-market.ppt, Assessed: 5/5/2003.
Stamper, R. K. 1992, 'Sings, Organisations, Norms and Information Systems', in Unknown (eds.), Proceedings of 3rd Australian Conference on Information Systems, Wollongong, pp 21-55
Swift, B. 2001, 'Making sense of application service providers, Part 1: Business computing comes full circle', Risk Management, vol. 48, no. 7, pp 21.
Tao, L. 2001, 'Shifting Paradigms with the Application Service Provider Model', Computer, vol. 34, no. 10, pp 32 - 39.
Taylor, F. W. 1911. The principles of scientific management, Harper & Bros., New York.
Tellis, W. 1997, 'Introduction to Case Study', The Qualitative Report, vol. 3, no. 2, pp.
Thorp, J. 1998. The Information Paradox. Realizing the Business Benefits of Information Technology, McGraw-Hill, New York.
Timbrell, G., N. M. Andrews and G. G. Gable 2001, 'Impediments to inter-firm transfer of best practice in an enterprise systems context', in Unknown (eds.), Proceedings of 7th Americas Conference on Information Systems, Boston, MA, pp 1084-1090.
van der Aalst, W. and A. Kumar 2001, 'A reference model for team-enabled workflow management systems', Data & Knowledge Engineering, vol. 38, no. 3, pp 335-363.
van der Aalst, W., A. H. M. ter Hofstede and M. Weske 2003, 'Business Process Management: A Survey', in M. Weske (eds.), Proceedings of International Conference on Business Process Management, Berlin, Springer-Verlag, pp 1-12.
Weijters, A. J. M. M. and W. M. P. van der Aalst 2002, 'Rediscovering Workflow Models from Event-Based Data.' in Unknown (eds.), Proceedings of Third International NAISO
Masters Thesis Chris Taylor
QUT Page 223
Symposium on Engineering of Intelligent Systems (EIS 2002), Sliedrecht, The Netherlands, NAISO Academic Press, pp 65-65.
Willcocks, L., D. Feeny and G. Islei 1997. Managing IT as a Strategic Resource, McGraw-Hill, London.
Williams, M. R. 1997. A History of Computing Technology, IEEE Computer Society, Los Alamitos.
Yao, Y. and L. Murphy 2001, 'Client Relationship Development for Application Service Providers: A Research Model', in Unknown (eds.), Proceedings of Proceedings of the 35th Annual Hawaii International Conference on System Sciences (HICSS-35.02), Hawaii, Computer Society, pp 3006 -3015.
Yin, R. K. 1994. Case Study Research: design and methods, Sage Publications Inc, Thousand Oaks California.
Zachman, J. A. 1987, 'A Framework for Information Systems Architecture', IBM Systems Journal, vol. 26, no. 3, pp.
Zamperoni, A. and P. Lohr-Richter 1993, 'Enhancing the quality of conceptual database specifications through Validations', in B. Thaleim (eds.), Proceedings of 12th International conference on the Entity-Relationship Approach, Berline, Springer, pp 85-98.
Zimmerman, E. W. 2002, 'Application service providers: Are the risks worth the rewards?' AFP Exchange, vol. 22, no. 1, pp 38.
Zwegers, A. J. R. 1998. On Systems Architecting – a study in shop floor control to determine architecting concepts and principles. Eindhoven, Eindhoven University.
Masters Thesis Chris Taylor
QUT Page 224
Appendices
Appendix A Quality Case Study Protocol
A study conducted by the Centre for Information Technology Innovation (CITI) QUT, Brisbane. Case Study protocol [design documentation and Investigators’ notes] Investigators
Chris Taylor Wasana Sedera CITI 2, George Street Brisbane QLD 4001 Tel 3864 9476 Fax 3864 1969 Email: [email protected]
School of Information Systems 2, George Street Brisbane QLD 4001 Tel 3864 1924 Fax 3864 1969 Email: [email protected]
Case study objectives To capture the ‘critical’ aspects that should exist in a good reference model To analyse the level of importance of these aspects To identify means of achieving these factors when applied in a real life context Case study design Broadly, the choice is between a single case study Vs a multiple case study approach. Both have there pros and cons, and issues that are specific to the context and resources of the study. We have selected a multiple case study approach. Unit of analysis Projects that use Reference process models Theoretical basis for Model Quality
Masters Thesis Chris Taylor
QUT Page 225
(adapted from Quality of Model Framework, (Lindland et al. 1994)) This Quality model will be used to examine and categorise the types of factors of a model that are important when using a reference model for Business Process Modelling projects. It was adapted from Lindland’s quality of Models framework and is presented here as the theoretical basis for the measures of what makes a model useful, in order to derive the list of critical factors a model must possess to be useful in a BPM exercise. Of particular interest in this research will be the aspects that exist in, or around, the model that assist with driving the quality of the model. A collective term “support” will be used to capture these aspects. The Support aspects are analogous to Lindland et al’s “means”. The framework above was used to help create the questions, by providing a framework that prompted questions into each area and as a classification tool for the questions (and their responses). This paper is based around the hypothesis that not all the different types of quality are important in all cases. The study is to examine which aspects of quality are important when using a business process reference model for business process modelling. Sampling frame Sampling parameters Possible choices Actors Reference Model users
Sponsors of the initiative using the reference models Process owners (who may already have fallen into any of the above categories)
Mode Individual Interviews Email correspondence Observations
Limitations:
Model
Language Domain
Audience
Masters Thesis Chris Taylor
QUT Page 226
Relying on respondent to separate perceptions from reality, esp wrt semantic vs pragmatic quality, e.g. if something was perceived as ‘wrong’ in the model was it a semantic issue (the model really is wrong) or a pragmatic issue (the model was interpreted incorrectly). “Reference models: critical success factors” Case study - Interview structure (Interview protocol) Compiled by Wasana Sedera and Chris Taylor State note of appreciation for time set aside for interview Get permission to use audio tape (if applicable)
1. Introduction 2. -State Overview 3. Study 4. Purpose of interview 5. Ethics 6. Context Information
Can you describe the overall goals of using <name of ref model> at <company name>? What motivated the organization to use <name of ref model>? Who was involved in making the decision and promoting the ref model? How did you learn about it? Is this a part of a larger initiative? Can you explain for what applications, you intended to use the ref models? How would you describe the overall approach you used to deploy the ref model at <X>? What were the main steps? (Please state and describe in detail) Ex: step 1: Project initiation - with users and modellers Training the modellers/users Etc.… How did you decide the scope of the project? How many models were developed overall (please specify, how many from each different level, if possible) Can we see some example models, please? Do you think that the scope of the project was realistically defined? Did you mange to finish this phase of the project in the estimated time frame? Did you develop models out side the initially specified domain? Were there any bottlenecks for resources? Where do you see further applications of <ref model name> at <X>? Have you done any formal ‘evaluation’ on this initiative? If yes, can you describe their evaluation process and the results of the procedure? Can you specify the benefits that <the ref model name> set forth for X, which would have otherwise been unattainable? Can you give an example, where <the ref model name> really helped? Do you believe that the organization has achieved the goals set forth for this project?
Masters Thesis Chris Taylor
QUT Page 227
What lessons (possible weaknesses) have you learnt from this exercise? (In other words, what would you do differently, next time?) Any specific project management concerns/ issues? Any specific modelling concerns/ issues? Testing the a-priori model What were the “issues” for you to proceed? How did you overcome them? (Collecting the negative aspect – of existing issues) What facilitated the task? (Collecting the positive aspect – of existing issue) Was there any thing <critical> that had to be there first, for you to proceed with the next steps? Is there anything that you wish, you had <upfront> to do a better task? How do you think this would have assisted the mapping initiative? (Collecting the aspects that the informant, may not have quite experienced, but would have appreciated to have) Syntactic Quality How well did the models stick to the pre-defined rules between shapes of the language? How well did the models stick to the pre-defined symbols/terms of the language? Did the model use them consistently?8 Did the models fit together according the rules (inter model)? 9 Were there any aspects of the language that had no correspondence to the domain? Were there any aspects of the language that had multiple or unclear meanings? Was the appearance of the models consistent with each other? (i.e. layout, naming conventions)? Where the rules applied consistently throughout the model10? Syntactical Support: What did, or would, you do to ensure syntactical correctness of any additions/modifications? Was there anything in the model (symbols, rules) that were not defined or explained? Did the model present a meta model11? Was a meta-model developed?12
8 from Becker
9 Reason: look at systematic design (GoM) layering, linking etc.BUT the consistence between different models is viewed as a part of semantic quality (zamperoni, lohr-richter 1993) in Becker 2000
10 similar to GoMs guideline of comparibility, i.e. that the rules have been applied identically, e.g. layout rules or naming conventions
11 question drawn from Becker, GoPM, for evaluation of the syntactical correctness of a
model it is indispensable to have an explicit (doeucmented) meta model)
12 This will indicate the importance of having a meta model
Masters Thesis Chris Taylor
QUT Page 228
Did the model provide help to ensure syntactical correctness on additions/modifications?13 What did you do to ensure syntactical correctness on additions/modifications? Did the model include any supplementary material (e.g. glossary, definitions)? Did you have a way of capturing information that was not supported by the language14? Did you use any other languages other than those used in the reference model? Semantic Quality How well did the model capture its area? Did the model cover all of its stated scope15? Were there many statements that were not relevant to you/the domain? Was there anything important/useful missing? Where there any errors in the model, i.e. things that would not be realistically possible? Was there anything in the model that you didn’t agree with? Were there many repetitions? Were there many trivial statements that were not beneficial? Where there many new ideas that would work well in real life (i.e. best practice)? Where there any contradictions in the model16? Semantic Support What did, or would, you do to ensure semantic correctness of any additions/modifications? Did you use any strategy to ensure semantic correctness on additions/modifications? Was any content fed back to the developers17? (have you got any examples?) One way to support Semantic quality is increasing Language-Domain fit: Were there any things that could not be properly shown in the model due to the modelling language? How did you handle this? Did you make any additions/changes to the symbols or rules18? (rules: within a model and across models) L-D Support Was the same model presented in different languages? Why was this done?
13 used to determine the need for a tool or a quality checklist for modified/created models
14 Both syntax and semantics
15 the scope determines the domain of the model
16 drawn from zamperoni 1993 “the first step to approach semantical correctness is to detect contradictions in the schema [i.e. model]
17 used to check the availability of a feedback mechanism for RM producers
18 could be both semantic and syntax
Masters Thesis Chris Taylor
QUT Page 229
Were different languages used to model different things? (E.g. data model for data and process flows for process) Pragmatic Quality Could the stakeholders understand the relevant parts of the model completely? Could the stakeholders understand the relevant parts of the model correctly? Could the stakeholders understand the relevant parts of the model easily? How often did the model get interpreted to say one thing later to find out that the model was actually saying something else? Did people disagree about the model?19 Was the level of abstraction correct? Was there any filtering (hiding of information) available? Was the model useful20? Pragmatic Support 21 Did the model include any supplementary material (e.g. implementation guides, procedural models) How did you ensure the model was laid out correctly (tool checklist guidelines? Were the models tailored to suit individual/groups users22? Does the tool/model support automatic tailoring of the models to suit audience23? Does the tool support simulation24? Were there any group discussions about the meaning of the model?25 Did the model include any supplementary material (e.g. extended explanations)? Did you do any training in the content as related to the model? Was the scope of the model clearly defined26? Comparability, inter model consistency of application of other guidelines. Was there a formal way of the breaking up/linking the models (hierarchy or division)27?
19 This could be a social aspect
20 More of a social perspective, but supported as pragmatic by shanks (print out) and FRISCO p145
21 things that don’t fall under the specific language, it is there to support the understanding and use
22 (Gulla and Brasethvik 2000) Pragmatic variation paradox: prag. Var. increases user comprehension, but complicates the formation of a common understanding of the domain.
23 (Gulla and Brasethvik 2000) [they said that despite tailoring the interpretations were consistent]
24 (Gulla and Brasethvik 2000) p
25 social level interaction
26 Based on FRISCO definition of pragmatics as “usefulness”
Masters Thesis Chris Taylor
QUT Page 230
Could you contact the developers of the model for clarification? How well did the area of the model overlap with your area? How easy was it to modify the reference models to suit the situation (e.g. update, configure, customize, modify)?28 How easily were the models displayed and distributed?29 Part of Pragmatic support is increasing the Audience-Language fit30: How well did the stakeholders understand the language used? How intuitive was the language to the audience? How easily could the language be used? A-L Support What training did the stakeholders have with the language used by the reference model? What experience did the stakeholders have with the language used by the reference model? Was there any explanatory material on the language used? Did you create any? Were there any changes to the language made to fit the audience? Part of Pragmatic Support is Audience-Domain fit: How familiar were the stakeholders with the area of the reference model in a general / (organizational) project specific sense? A-D Support What training did the stakeholders have with the area of the reference model? Based on this notion, do you think that the <name of ref model> was useful/ successful in this context?> (If yes, explain how?…if not, why not?) (Gulla and Brasethvik 2000)
27 Krogstie 95 some of the means to achieve pragmatic quality is structuredness, aesthetics
for diagram layout
28 the format in which the reference model comes will affect its easy of modification
29 associated with physical, from Gulla (Gulla and Brasethvik 2000) but This is really more of a physical level
30 Krogstie 1995 p225, actors familiarity with languages will effect their comprehension of a model in a specific langauge
Masters Thesis Chris Taylor
QUT Page 231
Appendix B Examples of classifications of Levels of
Reuse
The two models below were classified as having a high degree of reuse.
Figure 77: Example of High Level of Re-Use Classification (Incident Management)
The two model extracts below were classified as having a medium degree of
reuse.
Figure 78: Enterprise model example for medium re-use (Detect Incident)
incident detectedby client
incident detectedby support centre
staff
choosenotification method
phone methodchosen
email methodchosen
in person methodchosen
fax methodchosen
mail methodchosen
third party
client
phoneservice desk
sendan email
faxa letter
visithead office
sendmail
citecstaff
DetectIncident
RecordIncident
AssessIncident
ResolveIncident
CloseIncident
IncidentManagement
Enterprise model
1.1 DetectIncident
1.2 RecordIncident
1.3 AssessIncident
1.4 ResolveIncident
1.5 CloseIncident
1. IncidentManagement
Reference model
Masters Thesis Chris Taylor
QUT Page 232
Figure 79: Reference model example for medium re-use (Detect Incident)
The two model extracts below were classified as having a low degree of
reuse. Both were taken from the start of the Record Incident process.
Figure 80: Enterprise model example for low re-use (Record Incident)
Techniciandetects incident
Choose NotificationMethod
Phone methodchosenEmail method Chosen
Phone Help Desk
User detectsincident notin support
enabled application
RecordSymptoms
Webreportingchosen
RecordSymptoms
Service Staff
User User
User
User
Commentsentered
AssignClassification
DetermineResolution
ShiftOperator
IncidentKnowledge
ShiftOperator
IncidentKnowledge
IncidentResolutionnot known
Incident mustbe resolvedat Level 2
IncidentResolution
known
Incident isa ServiceRequest
SLA
Masters Thesis Chris Taylor
QUT Page 233
Figure 81: Reference model example of Low Level of Re-Use (Record Incident)
1.2.1 Create IncidentRecord
Incident RecordCreated from user
phone call
Incident RecordCreated from user
electronic correspondance
Incident RecordCreated from
Application report
Service Desk notified
Masters Thesis Chris Taylor
QUT Page 234
Appendix C Survey Forms Reference model use and impressions
Explanation: The “models” refers to the ARIS models (eEPC and Value Chain). The “books” refers to the textual descriptions [including the models included within the text]. Please put a N/A to questions that are not appropriate or that you cannot answer. Please try to answer all the questions. In the comparison fields please mark the box that indicates your choice. e.g. the following would indicate that your preference is blue, but it is not a strong favourite. Which is your favourite colour? Blue X Red PART 1: Demographics of Respondent How much modelling experience or training did you have before the project? (e.g. ITN252) Did you have any training or experience in IT services or Incident Management / Help Desk? How did you used the reference material (please tick any box/es as appropriate)?
Books Models Process Identification 3 4 AS-IS Modelling Guidance (template) 1 4 AS-IS Model analysis 3 3 TO-BE Modelling Guidance (template) 0 5 Performance metrics identification 1 3
Other: _____________________________
PART 2: Reference assessment The Books Was the use of the ITIL books helpful for the project? If yes, in what ways? Can you quantify the benefit in terms of time? What did you gain from the use of ITIL books? Which book(s) / chapter(s) did you use? What could be improved in the books? The Models Was the use of the process models helpful for the project? If yes, in what ways? Can you quantify the benefit in terms of time? What did you gain from the use of models? Which models did you use? What could be improved in the models? Direct Comparison Which contained more ambiguities? Books Models Which was quicker to understand? Books Models Which contained the most details? Books Models Which gave the best overview? Books Models Which one was easier to understand? Books Models
Which was more precise? Books Models Please select whether the books or the models are more likely to increase the following risks:
Masters Thesis Chris Taylor
QUT Page 235
Please rate which source was more helpful under the following criteria: Provides a basic solution, reducing the need for development from scratch.
Books Models
The best practice knowledge was accepted by the user/customer, efforts of convincing them was reduced
Books Models
The material allowed semantic errors to be detected in the AS-IS models. Books Models
The material allowed syntactic errors to be detected in the AS-IS models. Books Models
Ideas from the material can be easily incorporated into the TO-BE Books Models
The material was easy to discuss to reach an agreement on meaning Books Models
The relevance of the material could be easily determined. Books Models
The material helped to understand and use the modelling language Books Models
The material made it easy to understand the domain Books Models
Comparison of Use Which did you use most? Books Models Are the models are a good enhancement of the books content? y/n Future Use If I have to decide about either using the books or the models I would use the
• Books • Models
Why? Optimum benefit for similar projects in the future would be achieved through using
• nothing • the books only • the models only • the books and the models.
Other comments?
The developed TO-BE model is not the best for the specific situation Books Models
The time spent understanding the reference model was not worth the benefits gained
Books Models
Reliance on the reference model limited creativity and innovation Books Models
Masters Thesis Chris Taylor
QUT Page 236
Appendix D Detailed Survey Results How did you use the reference material Books Models
Process Identification 3 4 AS-IS Modelling Guidance (template) 1 4
AS-IS Model analysis 3 3 TO-BE Modelling Guidance (template) 0 5
Performance metrics identification 1 3
Other: _____________________________ 0 0
Did the models or the books: Average St Dev
contain more ambiguities? 3 1.4
quickest to understand? 3.4 1.5
contain the most details? 2.8 0.8
give the best overview? 4 0.7
make understanding easier? 4.2 0.8
provide more accuracy? 2.6 1.3
Which was more likely to increase the risk that:
The developed TO-BE model is not the best for the specific situation 3 1.5
The time spent understanding the material was not worth the benefits gained 2.2 1.0
Reliance on the reference model limited creativity and innovation 2.6 1.1
Where the models or the books more helpful in:
providing a basic solution, reducing the need for development from scratch. 2.8 1.6
allowing semantic errors to be detected in the AS-IS models. 4 1.2
allowing syntactic errors to be detected in the AS-IS models. 4 1.2
making it easy to incorporated ideas into the TO-BE 4.2 0.8
reaching an agreement on meaning 2.6 0.8
determining the relevance of the material 3.2 0.8
understanding and using the modelling language 3 1.5
making it easy to understand the domain 2.4 0.8
Which did you use most? 2.4 1.1 Table 33: Results of the Likert questions comparing the books and models
NOTE: 1 is the books while 5 is the models
Masters Thesis Chris Taylor
QUT Page 237
Appendix E Company Profiles31
Citec The fourth of the top ten IS outsourcing participants is Citec (#9) who
provided four participants at different times during the focus group sessions.
Citec is a government owned commercialised company stated in 1964 and
commercialised in 1992. Citec participants were managers with
responsibilities which ranged across the strategic management areas. They
brought a unique and valuable knowledge and experience taken from a
customer base of 57% government and 43% public clients. Citec have offices
in 5 Australian states and a 2001 revenue of AUD120 million and total staff of
700 plus people. Citec do not offer full service outsourcing, and focus on
integrated infrastructure management, e-Business solutions and application
outsourcing (Benson 2002).
Computer Science Corporation The participants CSC manage the IS needs for the worlds largest mining
company and have access to the collective knowledge of a multinational
organisation which has operated since 1959 in the United States and since
1970 in Australia. Knowledge management is a formalised process within
CSC and the participants were able to utilise this network of people and
databases in support of the research project. CSC has more than 4500
employees in Australia almost all of which work in the services market. IDC
report that CSC operates across the banking and financial services,
government and manufacturing industries (Benson 2002).
Corporate Service Agency Corporate Services Agency (CSA) was the only other government participant
in the research project. CSA was able to add to the input of Citec from the
view of a pure ASP with no hosting of applications. CSA was established in
July 1996 as a shared service provider of finance, human resource and
administrative services. The organisation is jointly owned by the Department
31 Many of these profiles are copied from Huxley’s masters thesis (Huxley 2003) with permission.
Masters Thesis Chris Taylor
QUT Page 238
of Natural Resources and Mines (NR&M) and the Department of Primary
Industries (DPI). The participant’s knowledge and experience stem from their
roles as managers of the SAP R/3 system delivered to NR&M and DPI along
with a number of legacy systems. CSA provides a full range of services to
support these systems and effectively operates as an Applications Service
Provider (ASP) to its clients. The agency’s revenue base is around $22M per
annum and there are approximately 260 full time equivalent staff (Corporate
Services Agency 2002).
Deloitte Touche Tohmatsu The two participants from Deloitte Touche Tohmatsu (DTT) were senior
managers within the Management Solutions group. They brought a wealth of
knowledge and experience to the research by providing insights gained from
their own experience but also that of the 259 Australian partners and 2554
professional staff. Like many of the large consulting firms DTT operates a
knowledge database on a global scale. The participants were thus able to
provide a global view of the industry. For the financial year ending June 2002
there was a Net Revenue of AUD573 Million.
The Management Solutions groups Deloitte also provides consulting
services, specialising in Business and Technology Strategy, Collaborative
Supply Chain Management, Strategic Sourcing and Procurement,
eMarketplaces, Systems Implementation, Process Transformation, Finance
Transformation, Human Resources and Change Management, CRM, and
Applications Integration (Deloitte Touche Tohmatsu 2002) .
EDS Consulting The participant from EDS operates at the operations end of the business
managing the takeover and set-up of new clients. This participant brings a
wealth of experience and knowledge to the construction of the physical and
informational infrastructure of complex outsourcing contracts. EDS is the
second largest IS outsourcing company in Australia with 26% of the IS
outsourcing market in revenue terms. In Australia the 2001 revenue reached
AUD 1.45 billion. Operations in Australia were established in 1986 and it has
some 7000 employees with operations in Banking and Finance Services,
Masters Thesis Chris Taylor
QUT Page 239
Communications, Government, the Manufacturing industry and
Transportation sectors (Benson 2002).
Hitachi Data Systems Hitachi Data Services (HDS) is a private wholly owned subsidiary of Hitachi
Ltd founded in 1989 (Japans largest electronics company). The participant
from HDS comes from a true niche player within the IS outsourcing industry.
This company specialises in providing remote data base services for all types
of applications (not just ES systems). This participant was able to provide
insights to the research gained from being a highly focussed specialist. As a
private company HDS do not publish revenue or employee figures, all
information about the company is consolidated in the group reports of the
Hitachi Corporation, Japan (Connected Corporation 2001).
IBM Global Services Australia The participant from IBM Global Services Australia (GSA) is a member of the
Global and Asia Pacific focus groups able to provide a global perspective as
well as understanding the uniqueness of the Asia Pacific and Australian
markets. IBM GSA has 10,500 employees in Australia with revenue of AUD
3.3 billion in 2001 of which AUD 1.725 billion was for Services. The company
operates mostly in the IS outsourcing and network infrastructure
management service fields competing in the Banking and Finance Services
sectors, Communications, Government, Manufacturing and Transportation
sectors. IBM Global Services support users in Australia, Singapore,
Malaysia, Thailand, Philippines, Indonesia, Hong Kong, Taiwan and Korea
from data centres in Sydney, Melbourne and Ballarat (Benson 2002; IBM
Global Services Australia 2002).
Mincom Mincom was founded in 1979 and is still majority owned by current and
former employees of Mincom (75.80%). The participant from Mincom brings
a specialised knowledge and experience in managing an outsourcing
business with enterprise system products produced by Mincom itself. This
was a unique perspective in the research project with no other participant
having such close and easy access to this type of knowledge. Mincom has
Masters Thesis Chris Taylor
QUT Page 240
over 1100 employees in 18 offices and operates in 40 countries. Revenue for
the 2001/2002 financial year was AUD 207.8 million. The company develops
and sells enterprise system software and provides IS outsourcing services for
its range of enterprise systems. It has a reputation for producing the world’s
best enterprise asset management systems for asset intensive industries
(such as utilities, mining and transport). (Hoover's Online 2002; IBIS World
2002; Paul Budde Communication 2002) (Mincom 2002).
Parmalat Pty Ltd The ninth participant and the only in-house organisation was Parmalat with
its head office in Milan Italy. The participant from their Brisbane headquarters
was able to provide the in-house view of enterprise system service provision.
Parmalat operate one of the latest versions of SAP R/3 in Australia with the
full suite of e-Business modules. They have a progressive and innovative IT
culture operating within the dairy industry which is a 24hours, 7 days a week
business. This drives the need for an IT service which is critical to product
management and thus a participant with a proven skills and knowledge in
developing effective and highly efficient processes within enterprise systems.
In 2001, Parmalat operated in 30 countries, with 146 plants employing
approximately 36, 000 employees and consolidated (group wide) sales of
AUD 13.5 billion (Parmalat 2001).
Queensland Rail Queensland Rail (QR) is a Queensland state Government owned corporation
that provides transport and logistics business solutions to a diverse range of
customers throughout the State, Australia and overseas. With annual
operating revenue of over $2 billion, 9500km of narrow gauge track, and
around 14,000 staff, QR is one of Australia’s largest and most modern rail
networks.
REALTECH AG The tenth company was the industry partner REALTECH and two members
of this company participated in the research. These participants were able to
provide valuable input from both the perspective of an outsourcing company
specialising in remote hardware and operating system maintenance and as a
Masters Thesis Chris Taylor
QUT Page 241
service provider to outsourcing companies as well. This enables the
participants to have a wide experience with the processes of many Australian
and international companies. The company itself is described earlier in the
chapter.
Telstra Telstra is majority owed by the Commonwealth Government of Australia after
going through partial privatisation in 1997, and provides information and
telecommunications services, throughout Australia and the Asia-Pacific.
Telstra’s revenue in 2002 was over $10 billion with over 17 million voice and
data services in operation employing over 40,000 staff.