IBM Industry Data Models · PDF fileIBM Industry Data Models ... EDW Architect – Asia...
-
Upload
truongtram -
Category
Documents
-
view
214 -
download
2
Transcript of IBM Industry Data Models · PDF fileIBM Industry Data Models ... EDW Architect – Asia...
®
IBM Software Group
© 2007 IBM Corporation
IBM Industry Data Models
Usage, Process & Demonstration
David Cope
EDW Architect – Asia Pacific
IBM Software Group
2
The EDW Data Model
� One of the essential components in the deployment
and consolidation of data warehouses and data
marts is the data model—the logical design
specification of the databases. Such Information
Models are the accepted way of designing all databases.
� Data models are critical path inputs to load process
(ETL) design. And a data model failure almost certainly results in lots of re-work:
� RE-WRITE ETL scripts and RE-RUN extractions
� RE-CREATE databases and RE-LOAD data
� RE-OPTIMIZE physical data base(s)
� RE-DESIGN and RE-BUILD OLAP applications
� RE-TESTall ETL and OLAP pplications
� RE-TRAIN end users
Planning
BusinessRequirements
Analysis
DataAnalysis
Data
Modeling
Design
Construction
Test andLoad
Training
IBM Software Group
3
IFW Components and Relationships
XSDM
XDWM
BSTs
XMLBridge
mapping
MetadataBridge
Services Data Model (XSDM)
Classification model for defining
business meaning across all
models and databases
Industry Data Model (XDWM)
Logical ER Model for designing central
data warehouse
Business Solution Templates (BSTs)
Logical Measures and Dimensions for
defining and presenting key performance
indicators (KPI's)
OSS/BSS Data Sources
OSS/BSS Data Sources
Logical E-R Models for
ODS source systems
mapping
ETTL
IBM’s solution is unique – there are a handful of Data Warehouse models available, but IBM’s solution is
accompanied by Model Management Tooling (m1) to utilize the power of the structure and deploy a full data classification across core business functions, further accelerated by fast-start Solution Templates.
IBM’s solution is unique – there are a handful of Data Warehouse models available, but IBM’s solution is
accompanied by Model Management Tooling (m1) to utilize the power of the structure and deploy a full data classification across core business functions, further accelerated by fast-start Solution Templates.
IBM Software Group
4
IFW Components in Physical Architecture
ROLAP
Business Definition Synchronization
CustomizedCustomized
BSTsBSTsMapMap
AggregationsAggregations
MOLAP
Sta
r Sc
he
ma
En
gin
ee
ring
Sta
r Sc
he
ma
En
gin
ee
ring
Scoped XSDMScoped XSDM’’
Data Mining
Mining Mining
VectorsVectors
Scores &Scores &
PropensitiesPropensities
DataWarehouse
DataWarehouse
Entity Na me
Fie ld -1 (PK)Fie ld -2 (FK)
Fie ld -3Fie ld -4
Fie ld -5Fie ld -6
Fie ld -8
Fie ld -9Fie ld -1 0
Entity Na me
Field- 1 (PK )
Field- 2 (FK )Field- 3
Field- 4Field- 5
Field- 6Field- 8
Field- 9
Field- 10
Entity Na me
Field- 1 (PK )Field- 2 (PK )
Entity N ameFie ld -1 (PK)
Fie ld -2
Fie ld -3Very Lo ng Field -4
Very Lo ng Field -5
Entity Na me
Field-1 (PK)Field-2
Field-3
Ve ry Lon g Fie ld-4Ve ry Lon g Fie ld-5
Entity N ame
Field -1 (PK)Field -2 (FK)
Field -3Field -4
Field -5Field -6
Entity Na meFie ld-1 (PK)
Fie ld-2 (FK)Fie ld-3
Fie ld-4Fie ld-5
Fie ld-6
Entity Na me
Field-1 (PK)Field-2 (FK)
Field-3 (FK)
Field- 4Field- 5
Field- 6
Entity N ameField -1 (PK)
Field -2 (FK)
Field -3 (FK)Fie ld -4
Fie ld -5Fie ld -6
Entity N am e
Fie ld -1 (PK)Fie ld -2 (FK)
Fie ld -3
Fie ld -4Fie ld -5
Fie ld -6
Entity Na meField-1 (PK)
Field-2 (FK)
Field-3Field-4
Field-5Field-6
Entity N am eFie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3
F ie ld -4Fie ld -5
Fie ld -6
Very Lo n g Fie ld -7Very Lo n g Fie ld -8
Entity Na me
Field-1 (PK)Field-2 (FK)
Field-3 F ield-4
Field-5
Field-6Ve ry Lon g Field-7
Ve ry Lon g Field-8
E ntity Na me
Field-1 (PK )
Field-2Field-3
Entity N am eFie ld -1 (PK)
Fie ld -2
Fie ld -3
Entity Na meFie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3 (FK)
Field-4Field-5
Field-6
Entity N am e
Field- 1 (PK )Field- 2 (FK )
Field- 3 (FK )Fie ld -4
Fie ld -5
Fie ld -6
Entity Na meField- 1 (PK )
Field- 2 (PK )
Entity N am e
Fie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3 (FK)
Class ifica tio n
Class ifica tio nClass ifica tio n
Class ifica tio n
Class ifica tio n
Class ifica tio n
Class ifica tio n
Cla ssific atio n
Cla ssific atio n
Cla ssific ation
Class ifica tio n
Class ifica tio n
Cla ssific ation
Entity Na me
Fie ld -1 (PK)Fie ld -2 (FK)
Fie ld -3Fie ld -4
Fie ld -5Fie ld -6
Fie ld -8
Fie ld -9Fie ld -1 0
Entity Na me
Field- 1 (PK )
Field- 2 (FK )Field- 3
Field- 4Field- 5
Field- 6Field- 8
Field- 9
Field- 10
Entity Na me
Field- 1 (PK )Field- 2 (PK )
Entity N ameFie ld -1 (PK)
Fie ld -2
Fie ld -3Very Lo ng Field -4
Very Lo ng Field -5
Entity Na me
Field-1 (PK)Field-2
Field-3
Ve ry Lon g Fie ld-4Ve ry Lon g Fie ld-5
Entity N ame
Field -1 (PK)Field -2 (FK)
Field -3Field -4
Field -5Field -6
Entity Na meFie ld-1 (PK)
Fie ld-2 (FK)Fie ld-3
Fie ld-4Fie ld-5
Fie ld-6
Entity Na me
Field-1 (PK)Field-2 (FK)
Field-3 (FK)
Field- 4Field- 5
Field- 6
Entity N ameField -1 (PK)
Field -2 (FK)
Field -3 (FK)Fie ld -4
Fie ld -5Fie ld -6
Entity Na me
Field-1 (PK)Field-2 (FK)
Field-3 (FK)
Field- 4Field- 5
Field- 6
Entity N ameField -1 (PK)
Field -2 (FK)
Field -3 (FK)Fie ld -4
Fie ld -5Fie ld -6
Entity N am e
Fie ld -1 (PK)Fie ld -2 (FK)
Fie ld -3
Fie ld -4Fie ld -5
Fie ld -6
Entity Na meField-1 (PK)
Field-2 (FK)
Field-3Field-4
Field-5Field-6
Entity N am eFie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3
F ie ld -4Fie ld -5
Fie ld -6
Very Lo n g Fie ld -7Very Lo n g Fie ld -8
Entity Na me
Field-1 (PK)Field-2 (FK)
Field-3 F ield-4
Field-5
Field-6Ve ry Lon g Field-7
Ve ry Lon g Field-8
E ntity Na me
Field-1 (PK )
Field-2Field-3
Entity N am eFie ld -1 (PK)
Fie ld -2
Fie ld -3
Entity Na meFie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3 (FK)
Field-4Field-5
Field-6
Entity N am e
Field- 1 (PK )Field- 2 (FK )
Field- 3 (FK )Fie ld -4
Fie ld -5
Fie ld -6
Entity Na meField- 1 (PK )
Field- 2 (PK )
Entity N am e
Fie ld -1 (PK)
Fie ld -2 (FK)Fie ld -3 (FK)
Class ifica tio n
Class ifica tio nClass ifica tio n
Class ifica tio nClass ifica tio n
Class ifica tio nClass ifica tio n
Class ifica tio nClass ifica tio n
Class ifica tio nClass ifica tio n
Cla ssific atio nCla ssific atio n
Cla ssific atio nCla ssific atio n
Cla ssific ationCla ssific ation
Class ifica tio nClass ifica tio n
Class ifica tio nClass ifica tio n
Cla ssific ation
ETLETL
ED
W E
RD
En
gin
ee
ring
ED
W E
RD
En
gin
ee
ring
ScopedScoped
XDWMXDWM’’
OE/OMOE/OMOE/OMOE/OMBillingBilling
OE/OMOE/OMOE/OMOE/OMCampaign MgmtCampaign Mgmt
OE/OMOE/OMOE/OMOE/OMCRMSCRMS
OE/OMOE/OMOE/OMOE/OMHRHR
OE/OMOE/OMOE/OMOE/OMCollectionsCollections
OE/OMOE/OMOE/OMOE/OMCommissionsCommissions
OE/OMOE/OMOE/OMOE/OMPOSPOS
OE/OMOE/OMOE/OMOE/OMA/P, A/R, G/LA/P, A/R, G/L
OE/OMOE/OMOE/OMOE/OM……
Legacy
OSS/BSS
IBM Software Group
5
� The XSDM is an object class hierarchy based on classification theory.
There are 5 levels in the TSDM: A, B, C, C’ and D
� There are 9 subject areas in the TSDM: InvolvedInvolved PartyParty, ArrangementArrangement,
Event, Event, ProductProduct, LocationLocation, ResourceResource ItemItem, ConditionCondition, ClassificationClassificationand BusinessBusiness DirectionDirection .
� The XSDM organizes these 9 data concepts first by Instance or Type and then as Fundamental, Associative and Attributive.
The Services Data Model (TSDM)
AA
BB
CC
CC’’
DD
Business SubjectsBusiness Subjects
Data ClassificationsData Classifications
Reusable EntitiesReusable Entities
Application EntitiesApplication Entities
Table StructuresTable Structures
IBM Software Group
6
The Nine Data Concepts in XSDM
• The 9 Data Concepts in the Services Data Model provide a generic overview of
the data in an enterprise
IBM Software Group
7
Industry Data Warehouse Model (XDWM)
INSTANCE TYPETYPE
ASSOCIATIVEASSOCIATIVE
FUNDAMENTALFUNDAMENTAL
ATTRIBUTIVEATTRIBUTIVE
InvolvedInvolved PartyParty
ArrangementArrangement
ClassificationClassification
LocationLocation
ResourceResource ItemItem
BusinessBusiness DirectionDirection
EventEvent
ProductProduct
ConditionCondition
IBM Software Group
8
XDWM Content and Structure – Concept
The set of Concept Fundamental (subtype) hierarchies will form the "skeleton" of an implemented ER model
All other data attaches to this skeleton
Known as Fundamental Entities
IBM Software Group
9
XDWM Content and Structure – Concept Descriptor
The set of Concept Descriptor (property) hierarchies will form the attributes of an implemented ER model
Also supplementary entities known as Attributive Entities e.g. repeated data
IBM Software Group
10
The set of Concept Relationship hierarchies will form the relationships of an implemented ER model
Also many-to-many entities known as Associative Entities
XDWM Content and Structure – Concept Relationship
IBM Software Group
11
XDW Implementation Approach
Conceptual Data Model
Workshops
Use JAD Sessions to customize
multiple
BST’s
BST’s
Express Requirements
in Enterprise Context Develop the framework for a cross-
functional Enterprise Information Architecture and specifies common
(and divergent) definitions of
business definitions
Defines Critical Information Elements and assists in the identification of
candidate data sources and change
data capture specification
Cross-Reference to Requirements
Expand Framework
into a Subject Area Specific Model
Information Framework
Physical Data Model
Cross-Reference to
Requirements and map to
XDWM
Logical Data Model
Refine design decisions and
expand the Physical Data
Model
Customize the XDWM to capture specific
requirements
Implement the PDM to support users analysis
requirements
IBM Software Group
12
Why use the IFW Solution
� Provides a Jump Start for analysis of warehousing requirements
� Includes Pre-Defined OLAP structures for analysis
� Pick and Choose Measures and Dimensions from Best Practise
� Has Consistent Structures and Common Definitions
� Is fully Customizable - scope required elements and add if required
� Completely Re-usable logical data model structures
� 100% Synchronized across XSDM, XDWM, BST & AST
� Supports Rapid Development
IBM Software Group
13
IFW DemonstrationConducted using the Banking Data Warehouse Model
IBM Software Group
14
IFW Comparison with other models
Limited or no software for model management, generation or
synchronization (i.e., model compare, merge or version control). No dedicated support for models and tools.
M1 tool provides an import/export bridge to RDA, ERWin and Essbase server as well
as model merge/compare and trace back functionalities. IBM provides dedicated software and helpdesk support.
Software Support
Deliverable on a wide-variety of platforms: hardware, OS, DBMS and ETL.
The highly normalized xDWM is easily
extended with new subtypes and attributes. Regular new versions covering industry developments and client feedback.
9 subject areas. 800+ entities and 50+ business solutions. Broader and deeper than other models.
The highly normalized xDWMconsolidates and integrates data from multiple data sources. De-normalized
summary tables and MOLAP cubes optimize query performance.
The IBM xSDM classifies and documents every entity, attribute, relationship and domain in the xDWM.
IBM IFW
Often completely proprietary based on the supported DBMS, OS or both.
Other models are basically one-off
solutions with no clear enhancement methodology. Not normalized enough to be extensible.
Data Models are not a core line of business for other vendors and R&D is often limited. Often subject areas and documentation are incomplete and there may be no business solutions.
Other models are either 3NF OR star schema-based but rarely both.
Other Models are not based on a classification model (taxonomy) or Information Framework.
Other Industry Models
Open Solution
Extensibility
Industry Coverage
Data Consolidation
and Scalability
Information
Framework
BENCHMARK
IBM Software Group
15