View
699
Download
2
Embed Size (px)
DESCRIPTION
Cloud blueprinting is a novel approach that lets developers easily syndicate, configure, and deploy virtual service-based application payloads on virtual machine and resource pools in the cloud.
Citation preview
Prof. Mike P. Papazoglou
Executive Director - ERISS Tilburg University, The Netherlands
email: [email protected], http://www.eriss.org
BLUEPRINTING SOLUTIONS for
CLOUD COMPUTING
European Research Institute in Service Science
• The ERISS vision focuses on how emerging technologies will impact organizations & society and provide critical insights to influence policy makers & businesses
Vision
Scientific Conferences, Programs & Summer Schools
• Founders of the flagship International Conference on Service Oriented Computing (ICSOC)
• Founders of the International Master in Service Engineering (IMSE)
• Joint founders of the highly successful Service and Software Architectures, Infrastructures and Engineering (SSAI&E) Summer School – supported by the EU FP-7 IST SSAI&E unit
Collaborative Research
• Joint research conducted with top research institutes in Europe, N. America, Australia & East Asia (China & Japan)
Expertise • Dedicated highly-qualified R&D team working at the intersection of business and technology
• Builds on a proven track record of envisioning the future, inventing and delivering the next wave of cutting-edge research solutions
Service Oriented Computing
Business Process Management
Cloud Computing
Research Expertise
Multidisciplinary
Research M
ethods
AGENDA
Overview, Vision & Aim Brief Introduc7on to Cloud Compu7ng The Cloud Delivery Model Landscape Managing the Cloud Blueprint Example: Interac7ve Telco Services Final Remarks
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia. 3
Overview, Vision & Aim
Vision: Smart Cloud Services
Smart traffic systems
Smart water management
Smart healthcare
Smart food systems
Smart supply chains
Smart cities
Smart business systems
Smart telephony
The world needs to get smarter – more instrumented, interconnected & lead to beLer decision making. Smarter Service & Cloud technologies are central to this vision.
(Process-‐intensive & event-‐driven applica3ons)
5 Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Aim of Talk
Describe an approach that supports the effec7ve deployment of global-‐reach service-‐based apps into a variety of different implementa7on plaPorms -‐ in par7cular federated cloud compu7ng forma7ons.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia. 6
Brief Introduction to Cloud Computing
Cloud: Consump7on & Delivery Models Op7mized by Workload
• A new consump7on and delivery model inspired by consumer Internet services.
Private, Public and Hybrid Workload and/or Programming Model Specific
The Industrializa7on of Delivery for IT supported Services
Cloud Services
Cloud Compu0ng Model
Infrastructure configuring Highly virtualized infrastructure Sourcing op7ons Economies-‐of-‐scale
Mul0ple Types of Clouds will co-‐exist:
“Cloud” represents:
Cloud enables: “Cloud” is:
Cloud Computing Overview
8 8 Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Cloud Computing Delivery & Deployment Models
Cloud Deployment
Models
9
Hybrid Private Public
Cloud service delivery model (provided by Cloud Providers)
Cloud Service
Applica0ons
Infrastructure as a Service (IaaS) • Virtualized servers • Memory, CPUs, Disk space • Networking
PlaEorm as a Service (PaaS) • Middleware – applica7on servers • Process automa7on middleware • Database servers • Provisioning, etc.
SoFware as a Service (SaaS) • Applica7ons (ERP, SCM, CRM) • Processes • Informa7on
Client-‐2 Client-‐n
Client-‐1
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Cloud Architecture
Standard APIs
Service Developer
Service Consumer
Standard APIs Service Provider
User Interface
SLAs/ contracts
Service Template Creation
Service Template Publication
Service Analytics &
Reporting Service Metering
Service Monitoring
Service Provisioning
Capacity Mgmt
SLA Mgmt
Service Billing
Service Reporting
Service Management
Hardware
Software Kernel (OS, Virtual Machine Manager)
Virtualized Resources
servers storage network
Infrastructure as a Service (IaaS)
Software as a Service (SaaS)
Cloud apps Cloud
apps
Cloud apps
Platform as a Service (PaaS)
Virtual Image Mgt
Image Library
Image
Data Privacy
Authentication &
Authorization
Auditing &
Accounting
Data Network Security
Certification &
compliance
Security
10 Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Purchasing
Billing & Collections Management
Service-based Application
Service Provider
Service
interface
Service-based Application
Service Provider
Service
interface
Service-based Application
Service
interface
Service-based Application
Service Provider Se
rvice
interface
End-to-end Logistics Processes
Inventory
SOA in the Cloud
11
Service Provider
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Cloud environment (platform & infrastructure providers)
Order Management
The Cloud Delivery Model Landscape
Inflexible Monolithic Cloud Delivery Models
• A monolithic one-size-fits-all SaaS/PaaS/IaaS stack architecture and vendor lock-in prevails.
• PaaS offerings are constrained by providers’
capabilities. They don’t allow easy extensibility, mashup, or customization options at the consumer or developer levels.
• Rigid service orchestration practices tied to a specific resource/infrastructure configuration at the application level.
• SaaS is predominantly tethered to proprietary application platforms in which the cloud provider runs all elements of the service and presents a complete application to the client. They’re hard to extend or customize.
Data
Runtime
Middelware
OS
Virtualization
Servers
Sotrage
Networking
Application
SaaS
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Example: Impediments of Current SaaS Solutions
-‐ Difficult to compose SaaS solu3ons in end-‐to-‐end processes -‐ Difficult to re-‐configure/customize SaaS solu3ons
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Stairway to the Clouds: Breaking the Cloud Delivery Monolith
15
S1
SaaS
PaaS
IaaS
S2
SaaS
PaaS
IaaS
Provider #1 Provider #2
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
The Syndicated Multi-channel Cloud Delivery Model
Benefits • Increased interoperability • Protec7on against vendor lock-‐in in the cloud
• Increased quality, callability, performance, (low latency, bandwidth)
• Control, reliability, simplicity, faster deployment, enabling new applica7ons
• New business models, innova7on, cost reduc7on
Virtualized applica0ons comprising end-‐to-‐end
Processes (BPaaS-‐Layer)
Client-‐2 Client-‐3 Client-‐n
SaaS-‐1 SaaS-‐2 SaaS-‐3 SaaS-‐n
PaaS-‐1 PaaS-‐2 PaaS-‐3 PaaS-‐n
IaaS-‐1 IaaS-‐2 IaaS-‐3
Applica0on
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Managing the Cloud
Meta-data Templates
• Meta-‐data templates: Templates describe how a cloud offering is presented & consumed. The offering is abstracted from the specific type of cloud resources offered. The provider uses service templates to describe in a general form what a cloud service can offer.
The Open Virtualiza7on Format (OVF) is an open, portable, &
flexible format for the packaging & distribu7on of virtual appliances (pre-‐built solu7ons comprised of one or more VMs).
By packaging virtual appliances in OVF, vendors can create a single, pre-‐packaged appliance that can run on customers’ virtualiza7on plaPorms of choice.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Model-driven Approaches
19
• IaaS model-‐driven approaches automate the deployment of complex IaaS services on cloud infrastructure.
A virtual appliance model treats virtual images as building blocks for IaaS composite solu7ons.
Virtual appliances are composed into a virtual solu7on model which helps developers determine deployment-‐7me requs in a cloud-‐independent manner using a parameterized deployment plan.
Composite appliances automate the deployment of complex app services on a cloud infrastructure.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
The Blueprint Cloud Delivery Model
The Blueprint Model
21
• The term "blueprint” refers to any detailed architectural plan, e.g., a technical drawing documen7ng an architecture or an engineering design.
• The blueprint model builds & documents a holis7c cloud architectural solu7on as assembled & operated, by engaging available cloud stack modules at all layers.
• It configures a unique op7mized cloud environment to meet specific a broad range of applica7on requirements & policies, which specify what is desired (e.g., consistency, security and privacy requirements) & not how it is to be accomplished.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
The Blueprint Model: Commoditizing the Cloud
22
SaaS Service
PaaS Service
IaaS Service
SaaS-BP
PaaS-BP
IaaS-BP
Blueprints Cloud Services
Describes
Describes
Describes
Provides abstract reusable and composable descrip7ons of cloud stack services that can be used as lego blocks to describe, create and deploy desired federated cloud architectures.
Features of the Blueprint Model
23
• It lets developers syndicate, configure, par77on & deploy virtual service app payloads on VM & resource pools in the cloud by clearly separa7ng service processing concerns.
• It alleviates vendor lock-‐in & promotes interoperability. – Any service can interoperate horizontally with another service at the same level of the cloud stack provided elsewhere.
– Upstream or downstream (ver7cal) interopera7on of cloud services is possible.
• It maps declara7ve configura7on points for abstract cloud service specs to available resources, & composes them into complete solu7on models (using simple aggrega7on and cross-‐configura7on of virtual services).
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Elements of the Blueprint Framework
24
Opera0onal service descrip0on, performance-‐oriented service capabili0es, resource u0liza0on
Query BDL & BCL templates & reason about correspondences,
mismatches, etc.
Based on model mgt algebraic ops, e.g., match, merge, compose, extract, delete, etc
Resource Le
verage
Any aspect of an SLA about a service: includes security, privacy,& compliance requirements
Blueprint Constraint Language
Blueprint Manipula0on Language
Blueprint Query Language & Reasoning Mechs
Declara0ve Blueprint Request Language
Blueprint Defini0on Language
Developer/user centric language
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
The Blueprint Cloud Lifecycle Model
Marketplace Repository
AD
Applica7on Developer
End user
Service provider
AD
SP
User
Customized source blueprints
Source Blueprint model
Customized source blueprints
Blueprint Query Engine
Blueprint Manipula0on Language
User
AD
+
Source Blueprint models
Interim Target Blueprint model
Deployment Plans & Configura0on Op0ons
Op0mized Target Blueprint model
SP Blueprint Defini0on Language
SP Blueprint Defini0on Language
SP Blueprint Defini0on Language
User Design
Design
Selec0on
25 Michael P. Papazoglou © "Research Roadmap in Service Oriented Computing" Summer School, Espoo, Finland - August 26, 2011
Cloud resources
Tes0ng &
Monitoring
Blueprint Defn Language: Specifying Blueprints
• Func7onal characteris7cs, including: • Service type • Messages • Interfaces • Opera7ons
• Defines the KPIs associated to the services, e.g.,: • Ranges of service availability • Latency • Bandwidth
• Describes the physical infrastructure & resources that are required to run the service described in the blueprint. Prototypical items include: average and peak workload requirements
• Policies: • Prescribe, constrain and specify any aspect of a
business agreement needed to use a service, including items such as security, privacy and compliance requirements.
Opera0onal Service Descrip0on
Resource U7liza7on
Policies Performance-‐oriented
Service Capabili0es
Virtual Resource Net
Blueprint Template
• Inter-‐connected abstract resources: states func7onal inter-‐dependencies & deployment dependencies & deployment op7ons
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Example: Specifying Vehicle Mgt Cloud Services <<Blueprint Vehicle Management App>>
• Operation Service Description • BlueprintID= VM-SaaS • Description=The vehicle management software provided by AutoInc • Ownership= {AutoInc, Software Industry, Netherlands, …} • Version= 1.0 • Release Date= 12/04/2011 • Capability= Vehicle Management Software • Service Signature
• Functionality = To manage the status and location of fleet vehicles • APIs Location = http//autoinc.com/apis • Endpoint Location =http://autoinc.com/endpoint • Range Nr Of Instances = {2,5}
• Resource Utilization • Resource Requirement= {‘AutoInc-Req01’, ‘JEE Application Server’, (1,1), ‘AutoInc-Req-QP01’}
• Resource Requirement= {‘AutoInc-Req03’, ‘MySQL DB’, (2,2), ‘AutoInc-Req-QP02’}
• ……
• Policy Section • Resource Constraint = Ethernet Exists Only” • Resource Constraint = “Synchronous Communication is absent ” • QoS Inv Profile = AutoInc-Inv-QP01 • Policy Inv Profile = AutoInc-Inv-PP01
<<Policy Profile >> AutoInc-Off-PP01
• Price/month = 10000 euros for <= 500 vehicles
• Used only in the Netherlands • All Dutch legal and tax issues are applied
<<QoS Profile>> AutoInc-Req-QP02
• Response time <= 3 s • Throughput >= 80 req/s
Example of a Virtual Resource Network
28
AutoInc-Blueprint
Directed Graph - Vertexes • Service Offering • Implementation Artefact • Resource Requirement - Edges • Abstract Resource Link • Horizontal Link • Vertical Link
VehicleMgt-SaaS
Data.zip App.war
2AutoInc-Req01 JEE App Server with Servlet Container -
PaaS
AutoInc-Req04 3Gbit Network Link
- IaaS
AutoInc-Req02 Relational Database
- PaaS
2
AutoInc-Req03 Context-as-a-Service- SaaS
- Horizontal Link: a functional dependency - Vertical Link : a deployment dependency - AbstractResourcekLink: connection to an abstract resource
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Blueprint Mgt Language: Unary Blueprint Operators
Operator Name
CreateBlueprint GetBlueprintByID GetBlueprintByProperties ModifyBlueprintProperties
GetProperty
DeleteBlueprint
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
BML - Binary Operators
Operator Name CompareBlueprintProperties
CompareBlueprints - Compare two blueprints
MergeBlueprintProperties - merge selected blueprint properties
SplitBlueprint - split a blueprint into smaller reusable blueprints
ComposeHorizontally - compose two or more blueprints at the same cloud stack level, e.g., SaaS to SaaS, PaaS to PaaS.
ComposeVertically - compose two or more blueprints at adjacent cloud stack levels, e.g., compose SaaS with PaaS, PaaS with IaaS. ResourceLink - Return the resources that will be claimed by this blueprint
Blueprint Constraint Language
Blueprint Predicates (BP)
Resource Utilization Constr.
SLA/QoS constraints
Example • FP1: ((Servlet 2.5 Container Exists ) Runs On (Intel Dual Core 2Ghz Exists))
• FP2-: Composition Engine )Exists
• FP4: Network Link 2Gbit Exists
• FP5-Link 2:N etwork Link 2Gbit Exists
• FP6-Link 3 2 Network Links 3Gbit Exists
Example • NFP1: Throughput >= 100 req/s
• NFP2: Availability >= 98% on 24/7
Security/Compliance const.
Example • RP1:use WS-Security & XML-Digital Signature
BCL is grounded on: • Linear Temporal and Monoidal t-‐norm based Logic (LTL/MTL)
• Compliance paLerns.
Deployment/Data residency constraints
Is composed of
Example • RP1: Only used in the Netherlands
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Blueprint Constraint Language
The CRL in BCL is designed for: • the formal specifica7on of compliance requirements. • enabling automa7c design-‐7me compliance verifica7on. • Grounded on:
• Temporal logic; Linear Temporal Logic (LTL/MTL) • Compliance paLerns.
• Supports non-‐monotonic requirements to relax rules and handle excep7onal situa7ons.
• CRL is an open extensible language.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Blueprint Example: Interactive Telco Services
34
Cloud Compu7ng and Rendering
• High quality of video & services – Bandwidth availability -‐higher priority to game traffic during network boLlenecks
– Video encoding is computa7onally demanding
• Low latency for interac7ve applica7ons – Real 7me new view rendering at the browser client end – Adap7ve Stream management to handle user requests and network loads
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
35
Interac7ve Mobile Video Gaming
Server
Internet
Video Encoding
Clients
Full-‐frame Rendering
CG applica7on
Out of resources!
Bandwidth: 2-‐6Mbit per client
CG app controls the shape, appearance, and mo7on of objects drawn using programmable graphics hardware.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innova7on”, 21-‐22 November, Wollongong, Australia.
Video Decoding
Video Decoding
Video Decoding
Gaming app Integrating Iaas/Paas/Saas Components
36
Image Cache
Data-‐base
Authen0ca-‐ 0on
Mobile Client
On line Rendering
Storage
Mobile Client
On line Rendering
Authen0ca-‐ 0on
Data-‐base
SaaS-‐view
PaaS-‐view IaaS-‐view
Image Cache Storage
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Blueprints for Interactive Video Application
37
Au * PlaPorm Descrip7on * -‐ API -‐ Endpoint -‐ Authen7ca7on conf.
* Infrastructure Descrip7on *
* Policies Descrip7on *
* Infrastructure Descrip7on *
* Policies Descrip7on *
* Constrained Infrastructure Descrip7on * − Medium size VM
* Policies Descrip7on * − Es7mate avg delay – if delay goes below 100 ms migrate to other IaaS − 99.999 % availability
* Applica7on Descrip7on * − Metrics: …
* Infrastructure Descrip7on * − Medium size VM − Medium storage
* Policies Descrip7on * − Deploy with rendering service − Use Amazon’s autoscale
* PlaPorm Descrip7on * − Beanstock
* PlaPorm Descrip7on * API -‐ Endpoint -‐ Database conf.
* Applica7on Descrip7on * − Metrics: delay, availability .. − Loca7on: URI − Version: … − Package: caching service − Environment: servlet
Authen0ca0on service BP
Rendering service BP
Caching service BP
Database service BP
Authen7ca7on Configura7on
Authen7ca7on Configura7on
DB Configura7on
DB Configura7on IaaS
Configura7on
-‐ Medium VM -‐ Large storage capacity
Check dependencies
Deploy
Deploy
Horizontal configura7on points Ver7cal configura7on points Exposure of horizontal parameters Externally enforced by
IaaS provider
DML composi0on point
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
Closing Remarks
Closing Remarks
39
• Contemporary cloud technologies are fraught with problems. New technologies are required to support the mgt of clouds and allow for the dynamic deployment and management of services.
• Blueprin7ng allows cloud crea7ng cloud forma7ons dynamically
to comprise an arbitrary assembly of virtual cloud services (business processes, virtual plaPorms, virtual machines, & virtual storage volumes) connected into whatever design of IT service-‐based applica7on & associated infrastructure a customer desires. – It provisions cloud services, effec7vely manages workload segmenta7on and portability, and supports cloud architectures that automa7cally manage the lifecycle of cloud services, par77on work & op7mize workload distribu7on.
Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.
References
1. M.P. Papazoglou and W.J. van den Heuvel, “Blueprin7ng the Cloud”, IEEE Internet Compu7ng, Nov/December 2011.
2. M.P. Papazoglou and M. Vaqueros “Knowledge-‐Intensive Cloud Services: Transforming the Cloud Service Stack”, in Knowledge Service Engineering Handbook, CRC Press, Taylor & Francis Group, 2012.
3. Distributed Management Task Force “Interoperable Clouds”, White Paper CIM Version 1.0, document DSP-‐IS010, November 2009, available from: hLp://www.dmP.org.
4. A. V. Konstan7nou et. al. “An Architecture for Virtual Solu7on Composi7on and Deployment in Infrastructure Clouds”, 3rd Interna7onal Workshop on Virtualiza7on Technologies in. Distributed Compu7ng, June 2009, Barcelona, Spain.
5. T.C. Chieu et. al. “Solu7on-‐based deployment of complex applica7on services on a Cloud”, Interna7onal Conference on Service Opera7ons and Logis7cs and Informa7cs, Qindao -‐ China, IEEE CS, August 2010.
6. F. Galan et. al., “Service Specifica7on in Cloud Environments Based on Extensions to Open Standards”, 4th Interna7onal Conference on Communica7on System Soxware and Middleware, June 2009, Dublin, Ireland.
7. A. Elgammal, O. Turetken, W.J. van den Heuvel, M.P. Papazoglou, “Root-‐Cause Analysis of Design-‐7me Compliance Viola7ons on the basis of Property PaLerns”, 8th Interna7onal Conference on Service-‐Oriented Compu7ng, San Francisco, December 2010.
41 Michael P. Papazoglou © Keynote: “Australian Symposium on Services Innovation”, 21-22 November, Wollongong, Australia.