integrating enterprise information resources - omg€¦ · full traceability, tool, ... mof...

Copyright 2000, UNISYS Corporation Slide 1

Integrating Enterprise Information Resources

(using OMG CWM, XMI and MOF)

OMG E-Business Integration WorkshopJanuary 29 2001 : Orlando, Florida

Sridhar IyengarUnisys Fellow

Member, OMG Architecture [email protected]

(949) 380-5692


Tutorial OverviewTutorial Overview

� EDOC, E-Business & Data Warehousing� OMG Modeling and Metadata Architecture

– Overview of MOF, XMI and UML

� E-Business Intelligence Architecture� CWM Overview� CWM Details� CWM and MDC OIM Status� Putting it all together


AcknowledgementsAcknowledgements

� Portions of this presentation have been derived from – OMG CWM presentation from the CWM team– My earlier tutorials/conference presentations

� Unisys corporation for the support and resources in defining and validating the CWM metamodel specification

� CWM team for their phenomenal work over the last 2 years!

� EDOC 2K for its ever expanding tutorial program


� Global services and technology company– 1999 Revenues : $7.54 Billion– www.unisys.com

� Focus on Enterprise Class Servers and E-Business Services around the world

� Actively influenced and using several OMG Technologies : UML, XMI, MOF, CWM etc.


OMG HistoryOMG History� 800+ Vendors and End User members� 1989 OMA Vision & Architecture� 1991 CORBA 1.0� 1995 CORBA 2.0 IIOP - CORBA Interoperability� 1997 MOF and UML adopted, Domain specs

begin to be adopted� 1999 XMI and CORBA Components adopted� 2000 CWM, XML/Value, EAI, EDOC, XMI for

XML Schema, Additional domain specs� 2001 UML for EDOC, UML 2.0, Better XML and

E-Business integration


The Data Integration ProblemThe Data Integration Problem(Cannot consistently use the data we have!)(Cannot consistently use the data we have!)

� Heterogeneity– Data Sources and Targets, Middleware, OS– Tools, Technologies

� Complexity of data– Structured (Largely relational)– Unstructured– Multidimensional

� Federated data sources and targets� Central Data Warehouse VS Subject Data Marts� Transformation� Systems Management etc. etc.


Solving the Integration ProblemSolving the Integration ProblemWhat is needed?What is needed?

� Methodology for building, evolving and integrating software� A mechanism for cataloging and indexing and searching

enterprise assets (metadata and data, internal and external)� A flexible service based distributed component architecture

that spans the enterprise� A set of shared standard vocabularies (information models)

and notation (meaning and context of data usage)� A metadata driven approach to automate integration between

islands of information� A set of open modeling, metadata and distributed computing

standards


Relevance to EDOCRelevance to EDOC

� Historically EDOC has focused on– Object and component middleware – Enterprise modeling and integration (including

metamodeling)

� But– Has not paid much attention to data integration and

database/warehouse middleware

� In reality– Data Integration is an ‘enterprise problem’– Data Integration on the Internet is even more painful– We can apply to ‘enterprise data’ some of the lessons from

‘enterprise objects’


Architect (XML!)

EE--‘‘Muddleware’Muddleware’ Architect’s Architect’s Dilemma Dilemma

BEAWeblogic

Distributed, Heterogeneous, Client/Server !Multiple Clients, Servers, Tools, Apps, O/S, File systems,

Databases, Repositories, Data Models, Object Models...

DataMartsSQL/Server

UnisysLINC,DMSII

WarehouseBuilder

Oracle Apps

MicrosoftSQL Server

SAPBI WareHouse

SiebelWarehouse

NCRTeradata

Middleware (Tuxedo, TIP, DCOM, IIOP, RMI, EJB, COM+,…)Information Models (MOF, UML, CWM, OIM, RSM, BODs, XML….)


OMG EOMG E--Business Integration Business Integration Bird’s eye ViewBird’s eye View

(C) Iyengar 2000

Distributed Runtime Middleware (IIOP, XML/Value, HTTP…)

E-BusinessApplication

DevelopmentUML, CCM

E-BusinessIntelligence

CWM

Information Models, Componentsand Metadata (XMI, MOF…)

E-BusinessApplicationIntegrationCCM, EJB,COM, XML

Community & Enterprise Information Portals (KM…)

Directory, Security, Database,Web, Transaction, Caching,

Metadata, Services..

HealthCare Financial Manufacturing Insurance...


The EThe E--Business Application Life Cycle Business Application Life Cycle and the Processand the Process

Architecture Centric, Business driven,Iterative and Incremental

And do this with quality in a distributed environment

Discovery and Transformation

BusinessRequirem

ents Modeling

Modeling Architecture,

Object, Data…

AddBusiness

Logic

Build /Wrap Components

Assemble& Test

Components

Configure &Deploy

Components

Models, Metadata, Components, Middleware

Manage

Components

ComponentRuntimes

(EJB, COM+)

Acquired

Components

WebTime!

Rigorous


Example Tool Integration ScenarioExample Tool Integration Scenario

Business,Componentand Data Modeling

Business,Componentand Data Modeling

Visual DevelopmentEnvironments

& Legacy Wrapping

Visual DevelopmentEnvironments

& Legacy Wrapping

EJB/COM+Generation &Deployment

EJB/COM+Generation &Deployment

Web App Srvror COM+

Web App Srvror COM+

ComponentManagementComponentManagement

JDBC Oracle Native

XMLXMLUMLUML

EJBEJB

CCMCCM

CWMCWMVCSVCS

MOFMOF

UREP

XML

Metadata @models,

components,code...Traceability

Links

WEBDAV*

SCC

MOF/JMI

XMI

Life cycle component ManagementFull Traceability, Tool, AppServer independent

XMI/UML

XMI/EJB

* In development




– Overview of MOF, XMI and UML� E-Business Intelligence Architecture� CWM Overview� CWM Details� CWM and MDC OIM Status� Putting it all together


Metadata Architecture EvolutionMetadata Architecture Evolution

COMPLEXITY

Hierarchical, MonolithicE/R ModelsProprietary InterchangeIBM, Burroughs

Data Administration

RelationalClient/ServerE/R ModelsCDIF InterchangePlatinum/CA, ViaSoft, MSP..

Y2000, Data Warehouse

OO, O-RRelational, XMLDistributed ObjectObject Models (UML,MOF)XML Interchange(XMI)Unisys, IBM, MSFT..

ComponentsFrameworksE-Business

1980 1990 2000 META MODEL EXTENSIBILITY


There is more & moreThere is more & more metadatametadatalurking everywhere!lurking everywhere!

Name Char

Marital Status Enumeration (M, D, W, S)

Age Integer

LastName String

First Initial String

Employee Number Number

REPOSITORY

Person Table

User::John Doe

Person Table Space

EmployeeRecord

John M 43

Mary D 27

Bill W 62

George S 18

Data

Meta Data

Meta Data

Meta Data

Created-by

Last good Archive 9/28/1998 : 2 AM

Archive


APIs APIs

APIs Defined

The Component BusCORBA/CCM; Java RMI/EJB; COM+; XML/XMI/HTTP

Object Adapter

TechnologyModels (I/F)

MOFXMIXML

UML, BO, CWM...

Distributed ComponentServices

Trading, Transactions, Agents...

ApplicationModels/Interfaces

Domain/BusinessModels (I/F)

Manufacturing, Finance...Business Object Initiative

Packaged ApplicationsTools

Component Repository

Component & Application Management

EE--Business Architectures : 2001Business Architectures : 2001


Object Services

MetaModels (UML, CCM, CWM…)

Tools, Applications, Databases, Registries

XML Metadata Interchange (XMI)Meta Object Facility (MOF)

Internet JavaCORBACORBA/COM Interworking

OMG Modeling and Metadata FrameworkOMG Modeling and Metadata Framework


OMG Modeling and Metadata OMG Modeling and Metadata FrameworkFramework

Model Driven App Development

B2B Application Integration

Data Warehousing

Vertical Industry SpecificationsUML

Model & Design

XMIMOF2XML DTDMOF2XML Doc

MOF2XML Schema*

JMI*MOF2Java* Coming

IDLMOF2IDL

MetaDataManagement

And So On

Framework Meta Manage


� The UML is a graphical language for– specifying– visualizing– constructing– documenting

the artifacts of software systems� Added to the list of OMG adopted technologies

in November 1997 as UML 1.1� Most recent minor revision is UML 1.3

(November 1999)

UML Quick TourUML Quick Tour


UML 1.3 : OverviewUML 1.3 : Overview

M o d e l_ M a n a g e m e n t(fro m U M L )

F o u n d a t io n(fro m U M L )

B e h a vio ra l_ E le m e n ts(fro m U M L )

U s e _ C a s e s(fro m B e h a vio ra l_ E le m e n t s )

S t a te _ M a c h in e s(fro m B e h a vio ra l_ E le m e n t s )

C o l l a b ora ti o ns(fro m B e h a vio ra l_ E le m e n t s )

C o m m o n _ B e h a vio r(fro m B e h a vio ra l_ E le m e n t s )

A c t iv it y _ G ra p h s( fro m B e h a v io ra l _E le m e n t s )

C o r e(fro m F o u n d a t io n )

D a t a _ Ty p e s(fro m F o u n d a t io n )

E x t en s i on _M e c h a n is m s

(fro m F o u n d a t io n )

Model businessprocesses and

behavior

Model businessdata/structures


OMGOMG Metamodeling LayersMetamodeling Layers

MIDDLEWARE

APPLICATION

MOF DTD/SchemaSchema of Schema!!

CWM, EAI, IDL, UML, DTD/SchemaMiddleware Schema

Business Models,DTDsApp Schemas

User ObjectsXML DocumentsBusiness data

XML Terms

<Acme_Software_Share 98789>, 654.56, sell_limit_order, <Stock_Quote_Svr 32123>

StockShare, askPrice, sellLimitOrder, StockQuoteServer[Models, Interfaces, BODs]

UML::Class,Attribute, CWM::Table, CWM::Cube[UML, CWM, EAI, IDL…]

MOF::Class, MOF::Attribute, MOF::Operation[MOF]

Sample Objects

User ObjectsLayer (M0)

ModelLayer(M1)

MetamodelLayer(M2)

Meta-Metamodel Layer (M3)

OMG TermsDefined By


MetamodelMetamodel ArchitectureArchitecture

<<metamodel>>MOF Meta-metamodel

<<metamodel>>UML Metamodel

User Model

<<instanceOf>>

<<instanceOf>>

<<instanceOf>>

:Foo

M3 Layer: Specifiesmeta-metaclasses forthe UML metamodel

M2 Layer: Specifiesmetaclasses for theUML metamodel, suchas Class

M1 Layer: Specifies classesfor the UML user models,such as Passenger, Ticket,TravelAgency

M0 Layer: User objects thatare instances of UML usermodel classes, such asinstances of Passenger,Ticket, TravelAgency

:Bar :Baz

<<instanceOf>><<instanceOf>>


MOF OverviewMOF Overview� Foundation for OMG Metadata and Modeling

architecture– Model, design and implement metamodels, (meta)model servers– Provides 4 generic meta-object interfaces for introspection (Used

by all MOF based Meta-models...)– Provides MOF-IDL mapping to automate generation of concrete

object interfaces for specific metamodels– Provides MOF-XML mappings to automate generation of XML

DTDs and Documents (XMI specification)– Provides MOF-Java (JMI*) mappings to automate generation of

Java classes and interfaces– Can also be used with COM using COM/CORBA Interoperability

software

� MOF uses UML for notation and design* JMI : Java Metadata Interface in development


M O F 1.1 Reflec t ive< < C O RB A ID L M odule> >

MO F M odel< < OM G M eta -m etam odel> >

M O F F ac ility

Discover & Manipulatemetadata

Model using UML Class Diagrams precisely

Find and ManageMetadata

Repositories

MetaMeta Object Facility (MOF) Object Facility (MOF) ArchitectureArchitecture


ReflectiveReflective module ofmodule of thethe MOFMOF

� All meta-models based onthe MOF inherit this package.

� This is the foundation for introspection in the OMG metadata architecture

RefBaseObject

meta_object()itself()repository_container()

<<Interface>>

RefPackage

get_class_ref()get_association()get_nested_package()

<<Interface>>RefObject

is_instance_of()create_instance()all_objects()value()set_value()add_value()add_value_before()add_value_at()modify_value()modify_value_at()remove_value()remove_value_at()invoke_operation()

<<Interface>>RefAssociation

link_exists()query()add_link()add_link_before()modify_link()remove_link()

<<Interface>>


OMG MOF 1.1/1.3 ModelOMG MOF 1.1/1.3 Model

ConstraintNamespace

Package

GeneralizableElement

0..*

0..*+supertype

{ordered}0..*

+subtype

0..*Generalizes

Classifier

Class AssociationDataType

Feature

BehavioralFeature StructuralFeature

Operation

AssociationEnd

Reference

0..*

1

+referent0..*

+referencedEnd1

RefersTo

Attribute

ModelElement

0..*0..1

+containedElement

{ordered}

0..*+container

0..1

Contains

0..*

1..*

0..*

+constrainedElement

1..*Constrains


� Use W3C Extensible Markup Language (XML) for the transfer syntax and interchange format– Specify XML Document Type Definitions (DTD) to enable

transfer and verification of • UML based models (using UML DTD)• Data Warehouse models (using CWM DTD)• Java and EJB artifacts (using Java and EJB DTDs)• MOF based metamodels and their instances (using MOF DTD)

� Specify a precise MOF to XML mapping– Allows interchange of any MOF based metamodel and

corresponding models (MOF--> XML Stream)– Enables automatic generation of XML documents that can be

validated with generated DTDs� Use UML and MOF for metamodel design and implementation� Designed by Unisys, IBM, Oracle, DSTC et al

XML Metadata Interchange (XMI)XML Metadata Interchange (XMI)


XML Streams (Models)(Many - based on each metamodel DTD)

XML DTD (MetaModels) (1 per metamodel used for validation)

CWMDTD

UMLDTD

JavaDTD

XMI 1.0, 1.1 SimplifiedXMI 1.0, 1.1 SimplifiedOMG Documents :OMG Documents : ad/99ad/99--1010--02, ad/9902, ad/99--1010--0303

XMLSyntax and Encoding

MOFMetadata Definitions

& Management

XMI

UMLMetamodel

Analysis & Design

UMLUML

Models UMLCWMModels UML

JavaClasses

Validate

XMI 1.0 : ad/98-10-05, 98-10-06


<!ELEMENT Class (name, visibility, isRoot, isLeaf, isAbstract, isActive,XMI.extension*,constraint*, requirement*, provision*, stereotype*,elementReference*, collaboration*, partition?,template?, view*, presentation*, namespace?, behavior*, binding?implementation*, generalization*, specialization*,parameter*, structuralFeature*, specification*,associationEnd*, participant*, createAction*,instance*, classifierRole*, realization*,classifierInState*, taggedValue*, ownedElement*,feature*)?>

<!ATTLIST Class XMI.element.att; XMI.link.att;><!ELEMENT name (#PCDATA | XMI.reference)*> <!ELEMENT feature (Feature| StructuralFeature| Attribute| BehavioralFeature|

Operation| Method Reception)*>...

UML 1.1 XML DTD SubsetUML 1.1 XML DTD Subset


Business

Customer

id : CustId

update()

A Simple Model A Simple Model

Let us pretend this is a trivial (meta)modelMOF and XMI work formodels and metamodels

<Model xmi.id="a1” name = “Business” visibility ="public">

<ownedElement><Class xmi. id="a7” name = “Customer” >

<feature><Attribute name = “id” >

<multiplicity><Multiplicity lower = “1” upper = “1”/></multiplicity>

<type>< DataType href=”|a247"/></type> </Attribute><Operation name = “update” scope ="instance"/></Operation>

</feature></Class>

</ownedElement></Model>

XMI 1.1 RepresentationXMI 1.1 Representation


XMI/MOF Tools XMI/MOF Tools Interoperability OptionsInteroperability Options

GeneratedInterfaces

for Meta(model)

Tool/App

MOF::ReflectiveReflective

Tool/App

Intermediate Stream/File (XML)

Model TransferDTD GenerationSchema Generation *(XMI)

MOF Introspection

ConcreteModel/Metadata Access

readwrite

XML

UMLCWM

ECEAI

* Coming in XMI


How to use MOF/XMIHow to use MOF/XMI

� Define the domain or technology specific model– Middleware models are usually called metamodels by OMG– Use UML - only knowledge of Class modeling needed to get started on

the design

� Export model to XMI processor which generates– XML DTDs for the specific (meta) model– XML documents that conform to the DTD

� Manage the Models, DTDs and documents in a MOF/XMI compliant distributed repository

� Use XMI toolkits from IBM, Unisys, DSTC and others� For generating federated (using CORBA) MOF servers use

toolkits from DSTC, Unisys etc.


XMI XMI -- Automobile Example Automobile Example (simplified)(simplified)

UML Model

AutoColor : StringDoor : IntegerEngine : Integer

<Auto><Color> Red </Color><Door> 4 </Door><Engine> 2 </Engine>

</Auto>

XMI Document

XMI

<!Element Auto(Color*,Door*,Engine*)>

XMI DTD, Schema

XM

I

MO

F

interface Auto{};

IDL, Java...

Class Auto{public String color;public int Door;public int Engine;}





� E-Business Intelligence Architecture� CWM Overview� CWM Details� CWM Status� MDC OIM Status� Putting it all together


Data and Metadata Discovery

Business ModelingERP Apps

ExtractCleanse

TransformLoad

Build Warehouse

Components

AssembleWarehouse

Components

Configure &Deploy

Mart/Whse

Database, Warehouse and Metadata Middleware

Manage

Warehouse

WarehouseMiddleware

Acquired

ComponentsData Sources

Data Warehousing / Business Data Warehousing / Business Intelligence Life CycleIntelligence Life Cycle

DomainModels,

Transforms

BI : Business Intelligence


The Federated Business Intelligence The Federated Business Intelligence ArchitectureArchitecture

� Grounded in the following technologies– Metamodeling architecture : OMG MOF– Metadata interfaces : OMG MOF IDL, JavaSoft JMI*– XML Metadata Interchange : OMG XMI, W3C XML– Data Warehousing Metamodel : OMG CWM– Data Access Interfaces : JDBC, ODBC, OLE/DB, JavaSoft

JOLAP*, Java for Data Mining*– Distributed Object and Data Access iddleware

� Designed by experts in databases, data warehouses, metadata management and distributed computing

* JMI and JOLAP are being defined as part of the Java Community Process


Firewall

Web ContainerEJB Container

Mainframe

CreditDW

JMI

JDBC

JMS

JDBCJSP

JSP

Servlet

Apps/GUI

EJB

EJB

EJB

EJB

Business Logic

Business

Object

Business

Object

Business

Object

Connector

ApplicationsIncome CreditRefinance HomeTarget Mailing Business Logic

Credit CheckDisposable IncomePurchase PatternsTarget Advertising

APPAppServer

ResourceAdapter

Security

Transaction

ConnectionMgmnt.

Client API

EIS

ClientsBusiness ObjectsCustomer ProfileAccount HistoryCredit HistoryLast MonthlyStatement

MetadataRepository w/ DW

Metamodels

OLTP DB

ApplicationDW

Java Business Intelligence ArchitectureJava Business Intelligence Architecture

(C) Sun Microsystems : Courtesy JavaOne JMI BOF

JOLAP

CWM





� E-Business Intelligence Architecture� CWM Overview� CWM Details� CWM Status� MDC OIM Status� Putting it all together


OMG Common WarehouseOMG Common WarehouseMetaModelMetaModel

� Scope– Data Warehouse lifecycle metadata interchange and

management

� Initial submission : IBM, Unisys, NCR, Hyperion, Oracle, Genesis, UBS, Dimension EDI...– Metamodel - Single logical & physical!– Generated XML DTDs– Generated MOF - IDL mappings– Generated XML document

� Approved by OMG Board in June 2000� CWM Finalization Task Force in Progress


CWM History CWM History -- HighlightsHighlights

� IBM, Oracle, and Unisys propose a Common Warehouse Metadata Interchange (CWMI) RFP to the OMG (6/10/98)

� CWMI RFP issued by the OMG (9/18/98)� Co-submitting team formed by IBM and Unisys, later joined

by Oracle, Hyperion, Genesis, NCR, UBS, and Dimension EDI (1/1/99)

� Liaison between OMG and MDC is formed following Microsoft submission of OIM to Meta Data Coalition

� An initial CWM Specification was jointly submitted by the team (9/17/99)

� CWM XMI Interoperability demo : 11/99� A final CWM Specification was jointly submitted for

evaluation and comments (2/11/00)


CWM History CWM History -- HighlightsHighlights

� The OMG ADTF votes unanimously to recommend adoption of the CWM Specification (3/10/00)

� The CWM FTF was formed to finalize the CWM Specification based on public comments and implementation feedback (3/10/00)

� The OMG Board adopts CWM (6/19/00)� CWM FTF (Chair Dan Chang, IBM) in process, completion

expected 11/00� CWM implementations underway at IBM, Oracle, Unisys,

Hyperion, SAS, MITI etc.


CWM Submitters & SupportersCWM Submitters & Supporters

� Submitters– IBM– Unisys– Oracle– NCR– Hyperion– UBS AG– Genesis Development– Dimension EDI

� CWM Specs: http://www.omg.org� CWM Info: http://www.cwmforum.org

� Supporters– Deere & Co.– Sun Microsystems– Hewlett-Packard– Data Access Technologies– InLine Software– Aonix– Hitachi


Source

Source

Source

Target

Target

Target

Source

Source

Source

CWM Tool

Target

Target

Target

Pairwise (9 connections)

Hub (6 connections)

Figure 1. Data warehouse source and target arrangements.

The CWM based Integration


CWMCWM� A complete specification of the syntax and

semantics needed to export/import shared warehouse metadata and the common warehouse metamodel, including:– The CWM Metamodel (Volume 1)– Interchange format for shared warehouse metadata

(CWM DTD, Volume 2)– Interchange format for the CWM Metamodel (CWM

XML, Volume 2)– Access API for shared warehouse metadata (CWM

IDL, Volume 2)

� http://www.omg.org/technology/cwm/


Roles of UML in CWMRoles of UML in CWM

� The metamodeling language (as in the MOF Model)– UML Semantics, UML Notation, OCL

� The foundation metamodel– UML Foundation, Common_Behavior, and

Model_Management packages

� The object (resource) metamodel– Same as above


CWM 1.0 Overview {02/2000}CWM 1.0 Overview {02/2000}Common Warehouse Metamodel Common Warehouse Metamodel

WarehouseProcess

WarehouseOperation

Transformation

XMLRecord-Oriented

MultiDimensionalRelational

BusinessInformation

SoftwareDeployment

UML 1.3(Foundation, Behavioral_Elements, Model_Management)

WarehouseManagement

Resources

Analysis

Object-Oriented

(UML)

Foundation

OLAP Data Mining

InformationVisualization

BusinessNomenclature

DataTypes Expressions Keys

IndexType

Mapping


CWM 1.0 Model CWM 1.0 Model -- Top LevelTop Level

CWMX

UML

CWM

The major packages in CWM:

org.omg.uml { UML 1.3 }org.omg.cwm { CWM Core }org.omg.cwmx { Extensions }


CWM Package CWM Package ArchitectureArchitecture

Modular Design– Minimum dependencies

• Cross package services provided by links to UML

– Avoid subpackages– Reduced complexity,

improved understanding– Use only the packages you

need

org.omgUMLCWM

Foundation<<metamodel>> DataTypes<<metamodel>> TypeMapping<<metamodel>> KeysIndexes<<metamodel>> Expressions<<metamodel>> BusinessInformation<<metamodel>> SoftwareDeployment

Resource<<metamodel>> Relational<<metamodel>> Record<<metamodel>> Multidimensional<<metamodel>> XML

Analysis<<metamodel>> Transformation<<metamodel>> Olap<<metamodel>> BusinessNomenclature<<metamodel>> DataMining<<metamodel>> InformationVisualization

Management<<metamodel>> WarehouseProcess<<metamodel>> WarehouseOperation

CWMX


CWM Extension Packages

Olap< <m etam odel> >

(from A nalys is )

Inform ationS et< <m etam odel> >

( from A nalys is )Inform ationReporting

< <m etam odel> >

(from A nalys is )

Record< <m etam odel> >

( from Resource)

M ultidim ens ional< <m etam odel> >

( from Resou rce)

COB OLData< <m etam odel> >

(from Resource)

DM S II< <m etam odel> >

(from Resource)

IM S Database< <m etam odel> >

(from Resource)

E ssbase< <m etam odel> >

(from Resource)E xpress

< <m etam odel> >

(from Resource)

DataTypes< <m etam odel> >

(from Foundation)

CORB A IDLDataTypes

< <m etam odel> >

(from Foundation)

UM L( from org.om g)

E R< <m etam odel> >

(from Foundation)

Inform at ionV isualization

< <m etam odel>

( from A nalys i s)

Extension Techniques•Metamodel specialization•XMI extensions•UML tagged values


CWM 1.0 CWM 1.0 -- Overview Overview

ResourceFoundation

AnalysisManagementER

<<metamodel>>

CORBAIDL<<metamodel>>


CWM 1.0 CWM 1.0 -- ResourcesResourcesRelational

<<metamodel>>

Record<<metamodel>>

Multidimensional<<metamodel>>

XML<<metamodel>>

IMSDatabase<<metamodel>>

COBOLData<<metamodel>>

DMSII<<metamodel>>

Essbase<<metamodel>>

Express<<metamodel>>


ViewisReadOnly : BooleancheckOption : BooleanqueryExpression : QueryExpression

SQLQueryquery : QueryExpression

Tableis System : Boolean/ optionScopeC olumn : Column/ type : SQLStr ucturedType/ usingTri gger : Tri gger

SQLSimpleTypecharacterMaximumLength : IntegercharacterOctetLength : IntegernumericPrecision : IntegernumericPrecisionRadix : IntegernumericScale : IntegerdateTimePrecision : Integer

SQLDistinctTypeleng th : Integerprecision : Integerscale : In teger/ sq lSimpleType : SQLSimpleT ype

1

0..*

sq lSimpleType

1

sqlDistinctType0..*DistinctTypeHasSimpleType

BaseTableisTemporary : BooleantemporaryScope : String/ trigger : Trigger

ColumnSet

SQLDataTypetypeNumber : Integer

CheckC ons traintdeferrability : DeferrabilityType

0..*

0..*

/constraint0..*

/constrainedElement0..*

{ordered}

Columnprecision : Integerscale : IntegerisNullable : NullableTypeleng th : IntegercollationName : StringcharacterSetName : String/ optionScopeTable : Table/ referencedTableType : SQLStructuredType

0..*

0..1 /feature

0..*{ordered}

/owner

0..1

0..*1/structuralFeature

0..* /type1

0..* 0.. *

/constraint0..* /constrainedElement 0.. *

{ordered}

CWM Relational Metamodel (Partial)


CWM 1.0 CWM 1.0 -- AnalysisAnalysis

Transformation<<metamodel>>

Olap<<metamodel>>

BusinessNomenclature

<<metamodel>>DataMining

<<metamodel>>

InformationVisualization<<metamodel>>

InformationSet<<metamodel>>

InformationReporting<<metamodel>>


CWM Transformation

� General mechanism describing data movement and lineage

� Generic transformations from any physical object to any other physical object

� Maps logical structures and concepts in the warehouse onto physical implementation

� Provides for multiple physical implementations of logical structures and concepts

Logical Object

Physical Object

Logical Object

Physical Object

Mapping Mapping

X-Form

X-Form

Conceptual Object

Conceptual Object

Mapping

X-Form

Mapping


CWM 1.0 CWM 1.0 -- Management Management

WarehouseOperation<<metamodel>>

WarehouseProcess<<metamodel>>

No Extensions


Catalog/Schema

Table ColumnRelational

Record RecordFile FieldRecordDef

XML Schema AttributeElementType

Multi-dimensional

Dimension DimensionedObject

Schema

ClassPackageObjectOriented

Attribute

RowSet Row ColumnValue

RecordSet FieldValue

Record

Document Element

Member MemberValue

MemberSet

ObjectExtent DataValue

Intent ExtentResource

DataValue

CWM Data Resource Matrix


CWM GenerationCWM Generation

<<metamodel>> CWM Packages(Rational Rose)

CWM Team

Apply UML to MOF Transformation Rules

Package XML Files

(MOF Model DTD)

MOF Facility

CWM PackagesCWM CORBA IDL

Files CWM DTD Files

CWMRepository

Server

IDL In/Out In/Out XMI

• Metamodel– Single logical source– Multiple packages

• Generated for each package – XML document – CWM DTD– CWM IDL

TheDefinitive

CWMMetaModel

Document TypeInterfacesUsed Unisys UREP/CIM forIDL and XML Generation


XMI for EXMI for E--Business Intelligence Business Intelligence November 1999 OMG DemoNovember 1999 OMG Demo

Distributed Runtime Middleware (IIOP, HTTP…)

E-BusinessApplication

DevelopmentUML, CCM

E-BusinessIntelligenceCWM, OIM

Information Models, Componentsand Metadata (XMI, MOF…)

E-BusinessApplicationIntegrationCCM, EJB,

COM+,XML

Community & Enterprise Information Portals (KM…)

Directory, Security, Database,Web, Transaction, Caching,

Metadata, Services..

HealthCare Financial Manufacturing Insurance...


CWM Case Study at UBS:CWM Case Study at UBS:The Business ProblemThe Business Problem

� UBS: One of the largest banks in the world� Legacy system needs to be migrated to new CBD-based

system� Legacy system uses

– COBOL record data models – Entity-Relationship models

� CBD system uses – UML object models – Different Entity-Relationship models


UBS: A CWMUBS: A CWM--Based Prototype Based Prototype SolutionSolution

� Use CWM to place legacy models COBOL record models and ER models in a repository

� Use CWM to place the new system’s object models and ER models in the repository

� Use CWM to define the transformations, and place models of the transformations in the repository– CWM’s transformation capabilities are key– A CWM transformation engine can execute the transformation without

additional programming by reading CWM-based transformation models

� UBS validated CWM by building repositories for these models


UBS Migration Repository: UBS Migration Repository: OverviewOverview

MigrationRepositoryPrototypeLegacy

System

C W M-Based COBOL record

structures

Transformation

C W M-Based Transformation

Models

U ML-Based Object Models

M OF (XMI, CORBA/ Java)



C W M-Based ER Models

C W M-Based ER Models

NewComponent

BasedSystem

(OMG, J2EE)

NewC W MBasedData

Warehouse

(Courtesy Hans-Peter Hoidn UBS)


Standardization Benefits for UBS❚ No lock-in to specific vendors❚ Models can be exchanged among different

vendors’ tools❚ Leverages OMG’s Meta Object Facility (MOF)

❚ Allows CWM data models and UML object models to be handled in an integrated fashioni(UML is also based on the MOF)

❚ MOF options for exchanging models among tools❚ XMI Leverages W3C XML standard❚ CORBA/Java


Roles of UML in CWMRoles of UML in CWM

� The metamodeling language (as in the MOF Model)– UML Semantics, UML Notation, OCL

� The foundation metamodel– UML Foundation, Common_Behavior, and

Model_Management packages

� The object (resource) metamodel– Same as above

CWM Metamodel Details : Part 1CWM Metamodel Details : Part 1

FoundationFoundationandand

Data ResourcesData Resources


Package ArchitecturePackage Architecture

Modular Design– Minimum dependencies

• Cross package services provided by links to UML

– Avoid subpackages– Reduced complexity, improved

understanding– Use only the packages you

need

org.omgUMLCWM

Foundation<<metamodel>> DataTypes<<metamodel>> TypeMapping<<metamodel>> KeysIndexes<<metamodel>> Expressions<<metamodel>> BusinessInformation<<metamodel>> SoftwareDeployment

Resource<<metamodel>> Relational<<metamodel>> Record<<metamodel>> Multidimensional<<metamodel>> XML

Analysis<<metamodel>> Transformation<<metamodel>> Olap<<metamodel>> BusinessNomenclature<<metamodel>> DataMining<<metamodel>> InformationVisualization

Management<<metamodel>> WarehouseProcess<<metamodel>> WarehouseOperation

CWMX


BaseBase

� UML notation used as diagramming technique

� UML metamodel extended to support warehouse concepts

The modeling environment

WarehouseProcess

WarehouseOperation

Transfor-mation

XMLRecord Multi-DimensionalRelational

BusinessInforma-

tion

SoftwareDeploy-

ment


Management

Resource

Analysis

Object(UML)

Foundation

OLAP Data Mining

InformationVisuali-zation

BusinessNomen-clature

DataTypesExpressions Keys

IndexType

Mapping


FoundationFoundation

Metamodels shared by other packages

� Foundation– Business Information– Data Types– Expressions– Keys & Indexes– Software Deployment– Type Mapping

WarehouseProcess

WarehouseOperation

Transfor-mation


BusinessInforma-

tion

SoftwareDeploy-

ment


Management

Resource

Analysis

Object(UML)

Foundation

OLAP Data Mining




IndexType

Mapping



� Business Information– Responsible parties & their contact information– Documentation and general commentary– Hierarchies of business types can be constructed

EmailemailAddress : StringemailType : String/ contact : Contact

LocationlocationType : Stringaddress : Stringcity : StringpostCode : Stringarea : Stringcountry : String/ contact : Contact

ResourceLocatorurl : String/ contact : Contact

TelephonephoneNumber : StringphoneType : String/ contact : Contact

Contact/ responsibleParty : Responsib leParty/ email : Email/ location : Location/ telephone : Telephone/ url : ResourceLocator

*

*

* {ordered}

*

*

*

*{ordered}

*

** url *

{ordered}

*

**

*{ordered}*

Documentreference : String/ modelElement : ModelElement

ResponsiblePartyresponsibility : String/ modelElement : ModelElement/ contact : Contact

*

*

*{ordered}

*

Descriptionbody : Stringlanguage : Name/ modelElement : ModelElement

ModelElement(from Core)*

*

*

*

*

*

*

*

*

*

*

*

Namespace(f rom Core)

*

0..1

+ownedElement *

+namespac e 0..1


FoundationFoundation� Data Types

– Supporting classes for modeling of data type systems– Extent class represents collections of instances

At tribute(from Core)

DataType(from Core)

ModelE lement(from Core)

Classifier(from Core)

TypeAlias/ type : Classifier

1

*

type1

*

EnumerationisOrdered : Boolean/ literal : EnumerationLiteral

EnumerationLiteralvalue : Expression/ enumeration : Enumeration

1

1..*

1

literal 1..*

StructuralFeature(from Core)

Union/ discriminator : St ructuralFeature

0..1

*

discriminator

0..1

*

UnionM embermemberCase : ExpressionisDefault : Boolean

0..1

*

/owner 0..1

/feature *{ordered}

Package(from M odel_Management)

Instance(from Common_Behavior)

Extent

*

0..1

/ownedElement *

/ namespace 0..1



� Keys & Indexes– Shared by several data resource models– Promotes similar representation across models

Relationsh ip(from Core)

ModelElement(from Core)

StructuralFeature(from Core)

KeyRelationship/ feature : S truc turalFeature/ uniqueK ey : Uni queKe y1..*

*

feature1..*

{ordered} *

UniqueKey/ feature : StructuralFeature/ keyRelationship : KeyRelationship

1..*

*

feature 1..*{ordered}

*

*

1

*

1

IndexedFeatureisAscending : Boolean/ index : Index/ feature : StructuralFeature

1

*

feature 1

*

Class(from Core)

0..1

*

/namespace

0..1

/ownedElement

*

0..1

*

/namespace0..1

/ownedE lement*Index

isUn ique : BooleanisSo rted : BooleanisPa rti tioning : Boolean/ indexedFeature : IndexedFeature/ spannedClass : Class

1

*

1

*

{ordered}

1

*

spannedClass 1

*



� Expressions– Tree-structured, functional model of expressions– Full access to all CWM objects– Supports both “black box” and “white box” expressions– Use “white box” expressions for interchange and lineage

ConstantNodevalue : Any Feature

(from Core)

Classi fier(f rom Core)

FeatureNode/ argument : ExpressionNode/ feature : Feature 1* 1*

ExpressionNodeexpression : Expression/ type : Classi fier/ featureNode : FeatureNode0..1 *

type

0..1 *

0..1

*

0..1

argument

* {orde red}


El ementNod e/ modelElement : ModelElement

1

*

1

*

Element(from Core)



� Software Deployment– SoftwareSystem

• A software packageon a CD

– A Deployed Software System is a set of DeployedComponents

• An installed SoftwareSystem

Subsystem(from M odel_M an agem ent)

Location(from BusinessInform ation)

Package(from Model_M anagem ent)

Si te/ containingSite : Site

*

*

containingSite

*

containedSite

*

SoftwareSystemsupplier : St ringversion : String/ types pace : Namespace

Component(from Core)

0..1

*

/namespace0..1

/ownedElement*

MachineinternetA ddress : String/ deployedComponent : DeployedComponent/ sit e : Si te

0..1

*

0..1

*

DeployedSoftwareSystem/ softwareSy stem : SoftwareS ystem

*1

deployment

*1

DeployedComponentpathname : String/ component : Component/ machine : Machine*1

deployment

*1

1

*

1

*

*

*

*

*

Node(f ro m Core)

– Each Deployed Component is on a specific Machine

• An installed program



� Software Deployment– DataManagers

provide access to data

• A deployed DBMS

– DataProviders wrap other DataManagers as specified by ProviderConnections

• ODBC and JDBC drivers

DataProvider/ resourceConnection : ProviderConnection

ProviderConnectionisReadOnly : Boolean/ dataProvider : DataProvider/ dataManager : DataManager

1

*

1

resourceConnection

*

Package(from Model_Management)

DataManagerisCaseSensitive : Boolean/ dataPackage : Package

1

*

1

clientConnection*

*

*

dataPackage *

*

Relationship(from Core)

Namespace(from Core)

SoftwareSystemsupplier : Stringversion : String/ typespace : Namespace

**

types pace

*

supportingSystem

*

Component(f rom Core)

0..1

*

/nam espace0..1

/ownedElement*

MachineinternetAddress : String/ deployedComponent : DeployedComponent/ site : Site

DeployedComponentpathname : String/ component : Component/ machine : Machine

*1

deployment

*1

1

*

1

*

Package(from M odel_Mana gement)

Node(f rom Core)


Re lat io nsh ip(from Core)

Pac kage(f rom Model_Management)


Ty peS ystemM appi ng

TypeMappingisB estMatch : BooleanisLossy : Bo ole an/ so urceTyp e : Classifier/ target Type : C lassifier

1

*

sourceType 1

mappingFrom*

1

*

targetType 1

mappingTo*

0.. 1

*

/namespace0.. 1

/ownedElement *

• Type Mapping– Map types to corresponding types in other systems– Designed for simple data type exchanges– Use Transformations for more complex mappings

Foundation


Data ResourcesData Resources

Describe logical and physical data containers

� Operational sources� Warehouse targets� Logical models

WarehouseProcess

WarehouseOperation

Transfor-mation


BusinessInforma-

tion

SoftwareDeploy-

ment


Management

Resource

Analysis

Object(UML)

Foundation

OLAP Data Mining




IndexType

Mapping


Data Resource MatrixData Resource Matrix

Catalog/Schema

Table ColumnRelational

Record RecordFile FieldRecordDef

XML Schema AttributeElementType

Multi-dimensional

Dimension DimensionedObject

Schema

ClassPackageObjectOriented

Attribute

RowSet Row ColumnValue

RecordSet FieldValue

Record

Document Element

Member MemberValue

MemberSet

ObjectExtent DataValue

Intent ExtentResource

DataValue


ViewisReadOnly : BooleancheckOption : BooleanqueryExpression : QueryExpression

SQLQueryquery : QueryExpression

Tableis System : Boo lean/ optionScopeC olumn : Co lumn/ type : SQLStr ucturedType/ usingTri gger : Tri gger

SQLSimpleTypecharacterMaximumLength : IntegercharacterOctetLength : IntegernumericPrecision : IntegernumericPrecisionRadix : IntegernumericScale : IntegerdateTimePrecision : Integer

SQLDistinctTypelength : Integerprecision : Integerscale : In teger/ sq lSimpleType : SQLSimpleT ype

1

0..*

sq lSimpleType

1

sq lDistinctType0..*DistinctTypeHasSimpleType

BaseTableisTemporary : BooleantemporaryScope : String/ trigger : Trigger

ColumnSet

SQLDataTypetypeNumber : Integer

CheckC ons traintdeferrability : DeferrabilityType

0..*

0..*

/constraint0..*

/constrainedElement0..*

{ordered}

Columnprecision : Integerscale : IntegerisNullable : NullableTypelength : IntegercollationName : StringcharacterSetName : String/ optionScopeTable : Table/ referencedTableType : SQLStructuredType

0..*

0..1 /feature

0..*{ordered}

/owner

0..1

0..*1/structuralFeature

0..* /type1

0..* 0.. *

/constraint0..* /constrainedElement 0.. *

{ordered}

Data Resource Packages• Relational

– RDBMS catalogs & ODBC/JDBC client catalog views– SQL-99 compliant


Mod elEle men t(from Core)

FixedOffsetFieldo ffset : In tegero ffsetUn itBi ts : Int eger

Package( from Model_Manag ement)

Attribute(from Core)

RecordFi leisSelfDescribing : BooleanskipRecords : Integer/ record : RecordDef

RecordDeffieldDel im iter : Str ingisFixe dWidt h : Bool eanreco rdDe l im ite r : St ringtext De l im ite r : Strin g/ fi le : Reco rdF i le

*

*

fi le*

record

*

{ordered}

Group

FieldValueConditionvalue : Expression/ field : Field

Field/ valueCondition : FieldValueCondition

*

1

valueCondition *

field1

Classi fier(from Core)

*

0.. 1

/feature *

{ordered}/owner

0.. 1

*

1

*

1

DataType(from Core)

Class(from Core)

/type

Data Resource Packages• Record

– Basis for traditional databases & files– Self-describing, delimited, & fixed-offset supported


MemberValue

Object(f ro m Co mmon _Beh avi or)

DataV alue(f rom Co m mo n_Beh avi or)

Extent(from DataType s)

Instance(from Com mon_Behavior)

0..1

*

/namespace0..1

/ownedElement

*

DimensionedObject/ dimension : Dimension

Schema

0.. 1

*

/namespace

0.. 1

/ownedElement *

Member

Dimension/ dimensionedObject : DimensionedObject/ component : Dimension/ composite : Dimension** **

{ordered}*

*

composite

*

component**

0.. 1

/ownedElement *

/namespace

0.. 1

MemberSet0..1*

/ namespace

0..1

/ownedElement

*

0..1

*

/nam espace

0..1

/ownedElement

*

Data Resource Packages• Multidimensional

– Physical representation of multidimensional databases


Schemaversion : StringxmlNamespace : String/ elementType : ModelElement

AttributedefaultKind : AttributeDefault/ elementType : Classifier

Contenttype : ContentTypeoccurrence : OccurrenceType/ elementType : ElementType/ ownedElementType : ElementTypeReference

ElementType/ schema : Namespace/ attribute : Feature/ content : Content

*

0..1 +/ownedElement

*+/namespace

0..1

*

0..1

+/feature *

+/owner 0..1

0..1

*

+content0..1

+elem entType*

Elem entContentorder : ElementOrderType/ ownedContent : ElementContent/ ownerContent : ElementContent

*

*

+ownedContent*

+ownerContent

*

ElementTypeReferenceoccurrence : OccurrenceType/ ownerContent : Content

Contenttype : ContentTypeoccurrence : OccurrenceType/ elementType : ElementType/ ownedElementType : ElementTypeReference

*

* +ownedElementType

*+ownerContent

*

Text/ ownerContent : MixedContent

MixedContent/ text : Text 1..11..1

+text

1..1

+ownerContent

1..1

ElementDocument *

0..1 +/ownedElement*+/namespace

0..1

Data Resource Packages• XML

– Supports XML 1.0– Basis for XML

documents– Allows use as

sources and as targets

CWM CWM MetamodelMetamodel, Part 2, Part 2

AnalysisAnalysis


Data AnalysisData AnalysisDescribes production & analysis of warehouses

� Describe analytical & deployment structures

� Design data movement & transformations

� Deployable on a number of data resources

WarehouseProcess

WarehouseOperation

Transfor-mation


BusinessInforma-

tion

SoftwareDeploy-

ment


Management

Resource

Analysis

Object(UML)

Foundation

OLAP Data Mining




IndexType

Mapping


Transformation� General mechanism describing data

movement and lineage

� Generic transformations from any physical object to any other physical object

� Maps logical structures and concepts in the warehouse onto physical implementation

� Provides for multiple physical implementations of logical structures and concepts

Logical Object

Physical Object

Logical Object

Physical Object

Mapping Mapping

X-Form

X-Form

Conceptual Object

Conceptual Object

Mapping

X-Form

Mapping


Transformation� Support for “White-Box”

transformation mappings

� Leverages the UML hierarchy: Mapping of Classifier-to-Classifier and Feature-to-Feature

� Mapping of Classifier-to-Feature

Feature(from Core)

TransformationMap/ classifierMap : ModelElement

Classi fier(from Core)

FeatureMapfunct ion : Proc edureExpressionfunct ionDescription : S tring/ source : Feature/ target : Feature/ classifierMap : ClassifierMap

1.. ** 1.. **

1..** 1..**

ClassifierMapfunction : ProcedureExpressionfunctionDescription : String/ source : Classifier/ target : Classifier/ transformationMap : Namespace/ featureMap : FeatureMap/ cfMap : ClassifierFeatureMap

*

0..1

*

0..1

1..**

source

1..**1..**

target

1..**

*

0..1

*

0..1

Class ifier(from Core)

Feature(from Core)

ClassifierFeatureMapfunction : ProcedureExpressionfunctionDescription : StringclassifierToFeature : Boolean/ classifier : Classifier/ feature : Feature/ classifierMap : ClassifierMap

*

0..1

*

0..1

1..** 1..**

1..** 1..**

source

target

classifier

feature


Transformation� Transformations can be

specified for arbitrary model elements

� Reification of the transformation “process”

� Relates to Warehouse Process and Warehouse Operation metamodels


DataObjectSet/ element : ModelElement/ sourceTransformation : Transformation/ targetTransformation : Transformation

1..**

1..**

Transformationfunction : ProcedureExpressionfunctionDescription : StringisPrimary : Boolean/ source : DataObjectSet/ target : DataObjectSet/ use : Dependency

* 1..**source

1..*

* 1..**target

1..*

TransformationTask/ t ransformation : Transformation/ inverseTask : TransformationTask/ originalTask : TransformationTask *

*

originalTask

*

inverseTask

**

1..*

*

1..*

TransformationAct ivit ycreationDate : String/ step : ModelElement

TransformationStep/ task : TransformationTask/ activity : Namespace/ precedence : Constraint/ precedingStep : Dependency/ succeedingStep : Dependency

*

1

*

1

0..1* 0..1*


OLAP� Analytical model: Cubes, measures, dimensions,

attributes, levels, and hierarchies� Cubes contain multiple measures and are implemented

via cube regions

CubeRegionisReadOnly : BooleanisFullyRealized : Boolean/ memberSelGrp : MemberSelGrp/ cube : Cube

MemberSelGrp/ memberSelection : MemberSelection/ cubeRegion : CubeRegion1 *1 *

CubeisVi rt ual : Boolean/ c ubeDim Assoc : CubeDimA ssoc/ c ubeRegion : CubeRegion

*

1

*

1

Hierarchy/ dim ension : Dimension/ cubeDim Assoc : CubeDim As soc

CubeDim Assoc/ dim ension : Dimension/ cube : Cube/ ca lcHierarchy : Hierarc hy

0..1

*

calcHierarchy0..1

*

*1 *1

MemberSelection/ dimension : Dimension/ memberSelGrp : MemberSelGrp1..** 1..**

Schema

0..1

*

0..1

* DimensionisTime : BooleanisMeas ure : Boolean/ hierarchy : Hierarchy/ memberSelection : MemberSelection/ cubeDimAssoc : CubeDimA ssoc/ displayDefaul t : Hierarchy

*

1

*

10.. 1

0..1

0.. 1displayDefault

0..1

* 1* 1

*

1

*

1

0..1*

0..1*


OLAP� Dimensions: Multiple

levels, attributes, and hierarchies

� Levels are used in multiple hierarchies; support a subset of the dimension attributes

Hierarchy/ dimension : Dimension/ cubeDimAssoc : CubeDimAssoc

DimensionisTime : BooleanisMeasure : Boolean/ hierarchy : Hierarchy/ memberSelection : MemberSelection/ cubeDimAssoc : CubeDimAssoc/ displayDefault : Hierarchy

1

*

1

*

0..1

0..1

displayDefault

0..1

defaultedDimens ion

0..1

LevelBasedHierarchy/ hierarchyLevelAssoc : HierarchyLevelAssoc

Level/ hierarchyLevelAssoc : HierarchyLevelAssoc

1

*

/dimension

1

/memberSelection*

HierarchyLevelAssoc/ levelBasedHierarchy : LevelBasedHierarchy/ currentLevel : Level/ structureMap : StructureMap/ listOfValues : StructureMap/ immediateParent : StructureMap

*

1

*{ordered}

1

*

1

*

currentLevel 1

StructureMap/ hierarchyLevelAssoc : HierarchyLevelAssoc/ valueBasedHierarchy : ValueBasedHierarchy0..1 *0..1 *

0..1

0..1

0..1 listOfValues

0..1

0..1

0..1

0..1 immediateParent

0..1

ValueBasedHierarchy/ structureMap : St ruc tureMap/ immediateParent : StructureMap

*

0..1

*

0..1

0..1

0..1

immediateParent0..1

0..1


Data Mining� Models the

fundamental meta data necessary for constructing and managing Data Mining models

� Three conceptual areas: Model, Settings and Attributes

MiningSettingsfunction : Stringalgorithm : String/ attributeUsage : AttributeUsageRelation/ dataSpecification : MiningDataSpecification

MiningModelResulttype : String/ model : MiningModel

MiningModelfunction : Stringalgorithm : String/ settings : MiningSettings/ inputSpec : ApplicationInputSpecification

0..1

1

settings 0..1

miningModel1 1

*

model1

*

ApplicationInputSpecification/ miningModel : MiningModel/ inputAttribute : ApplicationAttribute0..1 1..*

miningModel

0..1

inputSpec

1..*

ApplicationAttributeusageType : AttributeUsageattributeType : AttributeType

1..*

1

inputAttribute 1..*

inputSpec 1

SupervisedMiningModel/ target : ApplicationAttribute

1*

target

1*


Data Mining� Models major

settings types: Statistical, Clustering, Association Rules, Supervised

� Relates settings to specification and attributes

Sta tisticsSe tting s

ClusteringSettingsmaxNumberOfClusters : IntegerclusterIdAttributeName : String

RegressionSettingsCostM atrixClassificationSettings

/ costM atri x : CostMatrix0..1 *

costMatrix

0..1

se tt ing s

*

SupervisedMiningSettingsconfidenceAttributeName : StringpredictedAttributeName : StringcostFunction : String/ target : MiningAttribute

AssociationRulesSettingsminimumSupport : FloatminimumConfidence : Float/ i temId : MiningAttribute/ transactionId : MiningAttribute

Mi ningAtt ri bute

1

*

target

1

settings

*

1

*

itemId

1

settings*

1

*

transactionId

1

settings*

AttributeUsageRelationusageType : AttributeUsageincludeInApplyResult : Booleanweight : FloatsuppressNormalization : Boolean/ attribute : MiningAttribute

1

*

attribute1

a tt ri buteUsage

*

MiningDataSpecification/ attribute : MiningAttribute

1..*1

attribute

1..*

dataSpecification

1

MiningSettingsfunction : Stringalgori thm : String/ attributeUsage : AttributeUsageRelation/ dataSpecification : MiningDataSpecification

1..*

1

attributeUsage

1..*

settings1

1*

dataSpecification

1

settings

*


Information Visualization� CWM core metamodel

for information visualization and publishing

� Separation of “logical” rendered object from rendering “transformation”

� Recursive/composite structuring

XSLRendering


RenderedObjectformula : ExpressionNode/ rendering : Rendering/ defaultRendering : Rendering/ component : RenderedObject/ composite : RenderedObject/ neighbor : RenderedObject/ referencingNeighbor : RenderedObject

*

*

composite

*

component

*

*

0..1

*

0..1

** neighbor*

referencingNeighbor

*

RenderedObjectSet0..1

*

0..1

*

Renderingformula : ExpressionNode/ renderedObject : RenderedObject** **

0..1* defaultRendering

0..1*

0..1

*

0..1

*


Business Nomenclature� CWM metamodel for

“Business Metadata”� Intended for Data

Warehousing and Business Intelligence domains

� Common business terms and concepts

� Used in conjunction with analysis and reporting tools


Concept/ taxonomy : Namespace/ relatedConcept : Concept

Term/ glossary : Namespace/ concept : Concept/ relatedTerm : Term/ preferredTerm : Term/ narrowerTerm : Term

VocabularyElementdefinition : Stringexample : Stringusage : String/ modelElement : ModelElement


Business Nomenclature� Taxonomies consists

of concepts� Glossaries consist of

terms� Taxonomies and

Glossaries can be associated

� All are relevant to some “business domain”

BusinessDomain/ taxonomy : ModelElement

Taxonomy/ domain : Namespace/ concept : ModelElement/ subtaxonomy : Nomenclature

0..1

*

0..1

*

Glossarylanguage : String/ taxonomy : Taxonomy/ term : ModelElement/ subglossary : Nomenclature

*

*

*

*

Concept/ taxonomy : Namespace/ relatedConcept : Concept

*

*

+relatedConcept

*

*0..1 *0..1 *

Term/ glossary : Namespace/ concept : Concept/ relatedTerm : Term/ preferredTerm : Term/ narrowerTerm : Term

0..1 *0..1 *

*

*

*

*

*

**

+relatedTerm


Warehouse ManagementWarehouse Management

Orchestrates warehouse activities

� Warehouse Process relates– Transformations– Triggering events

� Warehouse Operation logs– Transformation

activity– Metrics

WarehouseProcess

WarehouseOperation

Transfor-mation


BusinessInforma-

tion

SoftwareDeploy-

ment


Management

Resource

Analysis

Object(UML)

Foundation

OLAP Data Mining




IndexType

Mapping


Warehouse ProcessWarehouse Process� A Warehouse

Process identifies warehouse tasks and the events that trigger them (“what gets done”)

� Relates Transformation Activity to Event types

ScheduleEvent

PointInTimeEv ent Interv alEv entduration : F loat

ExternalEv en tdescription : String

CascadeEv entwaitRule : WaitRuleTy pe

Retry Ev entwaitDuration : DurationmaxCount : Integer

RecurringPointInTimeEv entrecurringTy pe : RecurringTy pef requency Factor : Integer = 1month : Integerday Of Month : Integerday Of Week : Day Of Weekhour : Integerminute : Integersecond : Integer

CustomCalendarEv ent/ customCalendar : CustomCalendar

C ustomCalendar/ customCalendarEv ent : CustomCalendarEv ent

0..*

1

0..*

1

CalendarDatespecif icDate : Time

10..*

/namespace

1

/ ownedE lement

0..*

Event(from State_Machines)

Pac kage(from Model_Management)

WarehouseEvent/ warehouseProcess : WarehouseProcess

InternalEv entcondition : BooleanExpression/ triggeringWP : WarehouseProcess

Transf ormationActiv ity(from Transformation)

WarehouseProcessst at icD ef ini tion : Boolean = f als eis Sequential : Boolean/ transf ormationActivity : Trans format ionAct iv i ty/ int erna lEv ent : InternalEv en t/ warehouseEv ent : WarehouseEvent

1..*

1..*

1..*

1..*

0..*

1..*

0..*

triggeringWP

1..*

1..*0..*

1..*0..*



Warehouse OperationWarehouse Operation� Activity and Step

Execution track details of executions of Transformations (tracks “what got done when”)

� Step Execution may be related to a UMLCallAction model element

ModelE lement(from Core)

TransformationActivity(f rom Transformat ion) TransformationStep

(from Transformation)

ActivityExecution/ transformationActivit y : TransformationAct ivity/ stepEx ecution : StepExecut ion

1

*

1

execut ion*

CallAction(from Common_Behavior)

StepExecution/ transformationStep : TransformationStep/ activityExecution : ActivityExecution/ callAction : CallAction

*

1

execution*

1

0..1 *0..1 *

0..1*0..1*

Transformat ionExecutionstartDate : TimeendDate : TimeinProgress : Booleansuccessful : Booleanstatus : Expression


Warehouse OperationWarehouse Operation� ChangeRequest

objects represent a proposed change or one that has been implemented or rejected.

� Measurement objects can hold values for any object (such as volumetric details)

StepExecution/ transformationStep : TransformationStep/ activityExecution : ActivityExecution/ callAction : CallAction


Act ivityExecution/ transformationActivity : TransformationActivity/ stepExecution : StepExecution

Measurementvalue : Floatunit : St ringtype : StringcreationDate : Timeeffect iveDate : Time/ modelE lement : ModelElement

ChangeRequestchangeDescription : StringchangeReason : St ringstatus : Stringcompleted : BooleanrequestDate : TimecompletionDate : Time/ modelE lement : ModelElement

TransformationExecutionstartDate : TimeendDate : TimeinProgress : Booleansuccessful : Booleanstatus : Expression

CWM Generation, Validation, and CWM Generation, Validation, and ExtensionExtension


CWM Generation CWM Generation


CWM Team

Apply UML to MOF Transformation Rules

Package XML Files

(MOF Model DTD)

MOF Facility

CWM PackagesCWM CORBA IDL

Files CWM DTD Files

CWMRepository

Server

IDL In/Out In/Out XMIInterfaces

Document Type

� Metamodel– Single logical source– Multiple packages

� Generated for each package – XML document – CWM DTD– CWM IDL

TheDefinitive

CWMMetaModel


CWM Development CWM Development � Metamodel Proposal

– An information metamodel is proposed to be included in the submission

– Group analysis to determine if themetamodel should become part of CWM

� Model Team – A team is formed to develop a

metamodel as part of the CWM submission

InformationMetamodel for the

CWM team toconsider

CWM Team

Metamodel Accepted Metamodel Rejected

Information

Covered

Information

Outside Scope


CWM Development CWM Development � Metamodel Development

– Metamodels constructed by domain knowledgeable modelers

– A CWM Metamodel is constructed by analyzing common portions of existing Warehouse tool models as well as public reference models

– Tool logical models can be re-constructed using CWM as common starting point

Tool XMetamodel

Tool YMetamodel

Tool ZMetamodel

Common

Representation



CWM Validation CWM Validation � Use Case Scenarios

– Develop use cases of the individual models

– Develop representative use cases that involve several packages used in combination

– Develop use cases that represent entire target tools


CWM Validation CWM Validation Validation

– Each Use Case defines a “slice” through the model being validated

– Use Cases represent “typical” problems that the model must solve

OLAPModel

DeploymentModel

Abstraction Refinement

Semantic Mapping(Abstract) (Concrete)


CWM ExtensionsCWM Extensions

� CWM provides interchange of the common portions of warehouse toolmeta models

� CWM should be used as the new foundation of a warehouse tool model

� Volume 3 of the Specification contains examples of CWM Extensions

Tool XMetamodel

CWM

Metamodel


CWM Extension PackagesCWM Extension Packages

O lap< < m etam odel> >

(from A naly s is )

Inform ationS et< < m etam odel> >

( from A naly s is )Inform ationReporting

< < m etam odel> >

(from A naly s is )

Rec ord< < m etam odel> >

( from Res ource)

M ult idim ens ional< < m etam odel> >

( from Res ou rce)

CO B O LData< < m etam odel> >

(from Res ource)

DM S II< < m etam odel> >

(from Res ource)

IM S Databas e< < m etam odel> >

(from Res ource)

E s s base< < m etam odel> >

(from Res ource)E x press

< < m etam odel> >

(from Res ource)

DataTypes< < m etam odel> >

(from Foundation)

CO RB A IDLDataTypes

< < m etam odel> >

(from Foundation)

UM L( from org.om g)

E R< < m etam odel> >

(from Foundation)

Inform at ionV is ualization

< < m etam odel>

( from A naly s i s)


CWM ExtensionsCWM Extensions� CWM Extensions

– Extensions• Published tool

specific information for the purpose of interchange

• Physical characteristics of databases not common in general but interchangeable in a heterogeneous environment

Conjoint

Formula

Composite

Variable

Relation Dimension

ValueSet

AliasDimension


SimpleDimension

DimensionedObject(from Multidimensional)

Dimension(from Mul tidimensional)


CWM ExtensionsCWM Extensions� CWM Extensions

– Extensions• Common ancestry in

base metamodel• Tool-specific default

structures (e.g., Database)

OLAPServerApplication* ** *

PartitionDatabase

0..1

*

0..1

*

*0..1 *0..1

Outline

1

1

1

1

DimensionedObject(from Multidimensional)

Dimension

*

1

*

1

** **


Extending CWMExtending CWM� Extending the model to redefine a working tool metamodel without

generating a new DTD.– Categorized into three generic types:

• Proprietary Attributes: Tool specific definitions not intended for interchange

• Proprietary Associations and or Classes: Tool specific areas notcommon and not intended for interchange

• Sharable Extensions: Tool specific definitions proposed for interchange via tagged value pairs


Extending CWMExtending CWM� Extending the model to redefine a

working tool metamodel without generating a new DTD.


• Proprietary Associations and or Classes: Tool specific areas not common and not intended for interchange


CWM - Relational

Table

Column

ToolX - Relational

Column

DisplayName

ToolX - Relational.Column

Inherited Class Attributes

Leaf Class Attributes

InterchangedVia CWM

Proprietary







CWM - Relational

Table

Column

ToolX - Relational

Column

WidgetA

InterchangedVia CWM

Proprietary







CWM - Relational

Table

Column

ToolX - Relational

Column

DisplayName

ToolX - Relational.Column

Inherited Class Attributes

Leaf Class Attributes

InterchangedVia CWM

Interchanged Via CWM Tagged

ValuePairs

CWMCWMSummary and ActionsSummary and Actions


CWMCWM� A common specification that defines, in UML,

the structure and semantics of shared metadata in data warehousing and business intelligence– Resource: Object, Relational, Record, Multidimensional, XML– Analysis: Transformation, OLAP, Data Mining, Information

Visualization, Business Nomenclature

� A common specification that defines, in XML, the interchange format and, in IDL, the access API for such shared metadata


CWM Specification:CWM Specification:CWM XML, CWM DTD, CWM IDLCWM XML, CWM DTD, CWM IDLCW M

M etam odel(in UML

Notation)

CW M Specification

CW M M etadata Access

CW M IDL

CW M DTDCW M

Metadata Interchange

(in XM L)

CW M XML

MOF DTD


CWM Extensions (CWMX)CWM Extensions (CWMX)� Published vendor specific metamodel for the

purpose of metadata interchange (Volume 3 & Volume 4, non-normative)– Common ancestry in the CWM metamodel– Demonstrates the validity of the CWM metamodel– Demonstrates the extensibility of the CWM metamodel


CWM StatusCWM Status

� CWM FTF in progress since 3/2000� Weekly/bi-weekly teleconferences� Mail : [email protected]� Next revision 2/2001� Interoperability showcase scenario defined� CWM implementation in process� Preparations for incorporating MDC OIM

requirements underway– Following decision by MDC to continue metadata

standardization at OMG


� Microsoft joins MDC and transfers OIM (12/1998)� OMG and MDC have established liaisons with each

other– OMG/MDC joint meetings in 1999, 2000

� Some progress, efforts move in parallel– MDC already using OMG UML as a foundation for its Open

Information Model– OMG CWM spec uses OIM as design reference and

incorporates some features– OIM 1.x to incorporate some CWM enhancements– OMG XMI based DTDs for MDC OIM available

� MDC votes to continue metadata standardization at OMG (9/8/2000) instead of proceeding in parallel!– Work on unifying CWM and OIM begin 9/11/2000 at OMG

Burlingame meeting : ‘Unified Data Warehousing Standard’

MDC votes to build on OMG CWMMDC votes to build on OMG CWM


JSRJSR--40 : Java Metadata 40 : Java Metadata Interface (JMI)Interface (JMI)

� A new Java Community Process standardization efforts– Design well underway– Progress report at Java One in June 2000– Public Review planned for 4Q2000

� Builds on OMG MOF, XMI and UML specifications� Supports JSR-26 : UML-EJB Mapping specification being

developed in parallel� A pure Java metadata specification � Will be used in

– JSR-69 : Java for OLAP – JSR-73 : Java for Data Mining– JSR-26 : UML profile for EJB

� CWM to be used in the Java Business Intelligence platform


MOF, XMI for IntegrationMOF, XMI for Integration� Use UML for analysis and design of metamodels,

Information Model etc. (these are models after all!)� Define and manage metamodels/profiles using the MOF

– Use UML based modeling tools, or MOF interfaces (normative) or XMI

– Enables life cycle meta data interoperability and design reuse across metamodels

– Relationships and Subtyping across metamodels supported by the MOF

� Use XML for exchanging metadata via OMG XMI� Use MOF-IDL mappings for concrete IDL interfaces to

metamodels {JSR-40 Java Metadata Interface in future}� Inherit MOF Reflective interfaces for interoperable meta

objects across meta-models


Importance of (Importance of (MetaMeta)modeling using )modeling using MOF, XMI and UML MOF, XMI and UML

� Provides a regular and rigorous infrastructure at a higher level of abstraction

� Furnishes an architectural basis for extensions and evolution of software

� Facilitates alignment with other standards that use a metamodel architecture – Potentially eliminate or evolve redundant standards (eg: CDIF

now endorses XMI)� Supports interoperability and integration across

domains at the semantic level� Use UML to design metamodels and models, MOF

to implement and manage them and XMI to interchange them over the Internet


Importance of CWMImportance of CWM

� The industry standard for data warehouse interchange and interoperability

� Allows integration of enterprise information so that we can better leverage the data we have

� Already supports the most widely used data base and file systems

� Extensible to vendor/customer specific needs� Builds on UML, XMI and MOF� Accelerates data warehousing, portal and business

intelligence market


Who is Implementing XMI, Who is Implementing XMI, MOF, UML and CWM?MOF, UML and CWM?

� IBM VisualAge for Java, WebSphere, Rose tool kit� IBM VisualWarehouse, Hyperion, Oracle, SAS� Rational Software (Unisys XMI for Rose)� DSTC, OMEX...� Oracle Designer, Meta Integration, Together/J, Objecteering,

ObjectsByDesign, Aonix…� OMG standard metamodels and DTDs (MOF, UML, CCM,

CWM*, Java*, EJB*…)� Metadata Coalition voted overwhelmingly to build on OMG

specifications for metadata (September 2000)� Java Community Process JSR-40 : Java Metadata Interface*

* In progress


OMG UML/XMI/MOF UsageOMG UML/XMI/MOF Usage

UML MOF

XML

XMIUMLCWM

CCMEAI

EC

UMLCWMCCM

EAIEC

MOF CWMCCM

EvaluatingCORBA Med

UML

EJB Java UML Profile forEDOC...

IDLUML4CORBA

Submissions

Standards

Life Sciences

ElectronicCommerce

MDC OIM

Enterprise App Integration

DocumentManagement Etc.

SPE

CIAS


Concluding ThoughtsConcluding Thoughts� Ensuring a unified model driven distributed meta object

architecture is key to solving the heterogeneous integration problem

� Most customers have and will continue to have components and information/data from multiple sources and formats that need integration

� XML/HTTP, CORBA/EJB, DCOM/ActiveX, MOF/XMI (Metadata), UML (Modeling) and CWM (Data Warehousing) need to work together with the content (business models, BODS, HL7 RIM…)

� Modeling and Metadata matters - Master it : Use it in for your middleware, use it for your domains

� CWM is the key to integrating Enterprise Information


For More InformationFor More Information

� OMG : www.omg.org, www.omg.org/cwm� Unisys : www.unisys.com/marketplace/urep� UML : www.omg.org/uml, www.rational.com� CWM : www.cwmforum.org, www.omg.org/uml� W3C : www.w3c.org� DSTC : www.dstc.edu.au� Sridhar : [email protected]� UML RTF : [email protected]� MOF RTF : [email protected]� XMI RTF : [email protected]� CWM RTF : [email protected]� JMI : http://java.sun.com/aboutJava/communityprocess/jsr