Upload
eduardo-castro
View
733
Download
1
Embed Size (px)
DESCRIPTION
Citation preview
Dr. Eduardo Castro, MBA [email protected] http://ComunidadWindows.org
Empower Your
People
Self-Service
access &
insight
Familiar
interface &
tools you own
today
Improve
Organizational
Effectiveness
Alignment,
Accountability
& Ownership
Collaborative
across your
value chain
Enable IT
Efficiency
Trusted,
scalable &
secure
IT
Management
&
Interoperability
Low Total Cost of Ownership
Leverage your existing Infrastructure
Investment Strategy
Empower Your
People
Self-Service
access &
insight
Familiar
interface &
tools you own
today
Improve
Organizational
Effectiveness
Alignment,
Accountability
& Ownership
Collaborative
across your
value chain
Enable IT
Efficiency
Trusted,
scalable &
secure
IT
Management
&
Interoperability
Low Total Cost of Ownership
Leverage your existing Infrastructure
Investment Strategy
Empower Your
People
Self-Service
access &
insight
Familiar
interface &
tools you own
today
Improve
Organizational
Effectiveness
Alignment,
Accountability
& Ownership
Collaborative
across your
value chain
Enable IT
Efficiency
Trusted,
scalable &
secure
IT
Management
&
Interoperability
Low Total Cost of Ownership
Leverage your existing Infrastructure
Investment Strategy
Session Objective(s):
What is MDM?
Discuss Analytical MDM drivers
Understand BI/MDM process & data flow
Value proposition for MDM
How Microsoft MDM enables BI
MDM integration with SQL, IS, AS
8
ERP - USA
ERP - Europe
Data Warehouse or
Analytic App
•Difficulty coordinating multiple systems – Operational MDM
•Difficulty managing dimensions – Analytic MDM
•Or both – Enterprise MDM
Operational
•Europe keeps changing the color
•New products never include the
French product name
•I can’t pick the right Boeing
Analytic
•My sales reports are wrong
•Analysts can’t maintain dimensions
•Marketing can’t analyze by patch area
MD – Slowly changing reference data shared across systems
MDM – An application for creating and maintaining an authoritative view of master data including policies and procedures for access, update, modification, viewing between systems across the enterprise
PEOPLE
• Customers
• Vendors
• Sales people
• Employees
• Partners
• Patients
THINGS
• Products
• Organizational units
• Recipes
• Parts
• Storage bins
• Documents
PLACES
• Stores
• Wells
• Power lines
• Geographical areas
ABSTRACT
• Contracts
• Warrantees
• Time
• Metrics
Is this a real problem?
Tiger Brands has 23 different business units,
each using a different customer tracking
system
Microsoft had 9 “Boeing” top parent records
with 5907 different “Boeing” companies not
linked to any top parent records.
Compass Bank reduced the cost center
maintenance effort from 30 person days a
month to 2 by implementing a central
management process and system
Where is the real Boeing?
Purchasing
DB Asset Mgmt HR Doc
Boeing Boeing Inc Boeing
(Renton) Aut_Boeing Boein’
MDM
Boeing
Merge/Match
Versioning
Workflow Approval
Role Security
Hierarchy Mgmt
Master Data
Stewardship
Purchasing
DB Asset Mgmt HR Doc
Boeing Boeing Inc Boeing
(Renton) Aut_Boeing Boein’
15
Operational Analytic
•Any domain hub
•Stewardship portal
•Attribute association
•Business rules
•Workflow
•Email Notifications
•Granular security
•Model metadata
•Data quality tools
•Transaction logging
•Deployment tools
•Packaged content
•Hierarchy Management
•App-specific dimension builds
•Versioning
•Async Notifications
•App-specific integration
•Federation
Message
oriented -
BTS
Bulk
oriented -
SSIS
Office
Integration
SAP
Integration
“homeless”
data
Control
16
Customer Product
•Any domain hub
•Stewardship portal
•Attribute association
•Business rules
•Workflow
•Email Notifications
•Granular security
•Data quality tools
•Transaction logging
•Deployment tools
•Packaged content
•Model metadata
External
Services
Specific
Matching
Algorithms
Industry
standards
Federation
Complex data
model
Consistent
Schema
High data
volume
Diverse
Scenarios
17
SQL Imp.
Office Features &
Improvements
Stratature Base
New Features
•3 Tier UI
•Admin UI
•LOBI
•Metadata
•Model Deployment
•Powershell
•Services API
Product Adjustments
•Setup & Install
•Configuration Manager
•UI Updates
•Business Rules
•Authentication
•Model Deployment
•World Readiness
•Services API
•Security
•Documentation
•MetaData Explorer
•Workflow
•Performance
•Schema, Business Rule,
Granular Security, Testing
•Stress & Long Haul
•LOBi
•Async Export *
•Matching *
Improvements Completed
•Accessibility
•DB Schema Perf.
•Infragistics
•Internationalization
•Multi-Tenancy
•Security Config
•Setup
•SharePoint Int.
•UI Renovation
•ULS for MDM
•Windows Auth
Stratature Features
•Entity & Hierarchy Management
•Version Management
•Transaction History/Compliance
•Role-based, Granular Security
•Multi/any-Domain Models
•Central system of record
•Business Rules
•Notifications
•Bulk SQL Interfaces
85 – 95% Fact or Transaction
Data
5 to 15% Master Data
Master Data
•Ubiquitous
•Decays
•Reconciliation required
Fact Data
•Relatively Static
•Single Source
Master Data is Alive Needs Constant Management to Remain Valid
Decay
Data decays as the world changes around it
Corruption Data is corrupted as people and
processes touch it
Conflict
M&A and external relationships
• Deaths
• Mergers
• Marriages
• Divorce
• Bankruptcy
• Births
• Legislation
• Web Forms
• Mobile Forms
• Call Center
• Data Maintenance
• Data Input
• Internal Processing
• Inter- and Intra-company
• External Mandates (Wal-Mart / RFID)
• Multi-source Systems
• Federation
• Cross geography issues (reporting, rules)
• Data Silos
Problems multiply with scale and complexity • Number of geographies / governments
• Number of entities / customers / products
• Number of applications
Good BI = Good Facts + Good Dimensions
Dimensions define the structure for analysis Products
Customers
Centers – Revenue and Cost
Chart of Accounts
Fiscal Period
Hierarchies support consolidation and drill-down along dimensions
Good facts + bad dimensions = Bad BI
International and domestic accounting standards and legislation are demanding more visibility, transparency and audit ability of systems
Can you track changes in dimensions?
Do you know who made the changes?
Do you know when changes occurred?
Can you produce a dimension from Q2 last year?
Most analytics tools use a single administrator
Few if any have version control
Customers often believe they can have a single homogenous DW or ERP that provides all data with all dimensional categories needed for analysis. Frank conversations usually reveal that this is not realistic
The most important Master Data is often “homeless”
How do you map multiple customer IDs into one recognized relationship?
Your ERP has a product group but how do you reflect other important product attributes and hierarchies?
How do you exclude the effect of acquisitions?
How do you merge multiple chart of accounts into a common structure?
How can you analyze customers by market?
Data Stewardship Support
• Application Integration
• User transparency
• Approval and Notification
• Manual Data Conflict Resolution and Reconciliation
• Data Steward Console
Data Browsing and Editing
• Hierarchy Management
• Entity Management
• Role based and Context sensitive browsing
• Ubiquitous integration
• Arbitrarily defined views
Change Management
• Versioning
• Histories
• Slowly Changing Dimensions
• Slowly Changing Hierarchies
Data Rules Engine
• Customizable
• Pre-configurable
• Extensible
• Integrated with workflow
• Factors in authoritativeness
Security
• Context Sensitive Entity Views
• Value and Role Based Security
• Security based on Metadata and attributes
• Transaction logging and rollback
Master Data Synchronization
• Write-back
• Real time updates
• Automatic Data Conflict Resolution and Reconciliation
• Bulk Import/Export into hub
Master Data Metadata
• Entity tagging
• Entity prioritization
• Lineage
• Impact Analysis
• Referential Tracking
• Descriptions, Usages, Custom
High Speed Merge/Match Engine
• Ad-hoc real-time lookups
• Bulk capabilities
• Third party data integration
• Diff/Update capable
• Cross mode results compatible
Enterprise integration
• Thin client capable
• Office thick client available
• SharePoint integration
• Easier on-boarding
• Third party application integration
• Federated aggregation
Process agnostic central repository for master data that is highly available to organization
Allows us to build multiple dimensions from complete representation of the master data from department or organization
Web based UI and security model enables users to have very limited update permissions without providing access to sensitive data.
Server based data provides organization with access to master data 24/7
AS is great at BI analytics
Cross dimension analysis
Slicing, dicing, aggregations
Usually requires a cube context
AS not good at managing change
Core engine not optimized for edits
Write-back possible, but difficult to audit, manage & surface
No workflow, notifications, business rules, auditing, stewardship UI, fuzzy matching
MDS provides the ability to manage hierarchies, using explicitly (arbitrary) relationships or derived from data Provide attributes and attribute groups at parent and child nodes Slowly changing dimension semantics Allows for the automatic and accurate creation of hierarchy paths based on attribute relationships Enforces the structure and security of the hierarchy by preventing inaccurate relationship assignments This provides business intelligence users with a wide array of possible dimensions
Benefits
Manage multiple versions without cluttering individual BI application ie AS, PPS , TM1, Cognos Planning
Version flag
Support effective dates
Analysis Services
Permit multiple snapshots of dimensions
Used for VSCD (very slowly changing dims)
Customer2007, Customer 2008, Customer
Benefits
We provide robust business rule engine and workflow to maintain data quality for dimensions.
Transaction Management & Auditing
Analysis Services
Clean up the data presented to the business
Flag problems in records
Automatically categorize entities
SQL Server Analysis Services
SQL Server Integration Services
Excel & SQL Server Reporting Services
Planning Applications
LOBI/BDC Integration
Traditional BI SSIS loads warehouse AS loaded from WH MDS from WH Dimensions
Mastered BI
SSIS loads MDS WH Dimensions loaded from MDS AS loaded from WH
Source SSIS DW AS & MDS
Source SSIS MDS WH AS
demo
Central management of dimensions Product, Cost Center, Chart of Accounts, Customer, others….
Audit-ability / Trace-ability Tracking of changes by user, time and other change details
Version control Preservation of past dimension versions and comparison highlights among versions Promotion of new versions to subscribing systems and notification of version changes to users
Integration with OLAP, ERP and data warehouse
Preguntas
Ing. Eduardo Castro, PhD [email protected]