Technical vs. Business Metadata

1

A Standards Based Architecture for Federated
Data Management in Support of the FEA DRM
4/30/03
2
A Best Practice for E-Government

• Simplification should start with unifying access
  to data stores.

How?
3
Topics

• Metadata and MOF
• Model Driven Integration
• Example Use Cases
• The Federal Enterprise Architecture (FEA) Data
  Reference Model (DRM)
• The New Department of Homeland Security

4
The Relevant Standards

• ANSI SQL
• Java J2EE JDBC, JMS, JNDI, EJB, RMI
• XML W3C XML, XML Schema, SOAP
• OMG MOF, XMI (MetaMatrix is a member of OMGs
  CWM, XMI and MOF task forces)

Investment Protection and Interoperability
5
Federated Data Management

• Must be a standards-based approach
• A Web services approach is the best candidate for
  data interchange
• Steps
• Define XML Schema
• Map Schema to physical data stores
• Define WSDL
• Get XML document
• Seems simple, straightforward
• However, the mapping of XML schemas to disparate
  physical data stores can be a black hole

6
What Every CIO Needs to Know About Metadata

• Quotes from a CIO recommendation paper written by
  the federal government CIO council's
  interoperability committee
• Metadata is one of the biggest critical success
  factors to sharing information and storing
  information cost-effectively. Metadata can make
  your information sharing and storage efforts
  great successes, or great failures.
• Simply put, metadata management is making
  metadata do more things for more people in more
  ways, more economically.
• The alternative to metadata management is
  information chaos.
• http//www.cio.Gov/documents/metadata_cio_knowledg
  e_feb_1999.html

7
Federated Data Management - Metadata

• Federated Data Management is an enormous
  challenge.
• Understanding the underlying metadata must take
  place before any integration efforts begin.
• Post-it notes
• Spreadsheets
• Databases
• Registries
• ISO 11179, data element specific, datatypes,
  attributes
• ebXML, higher level organization, structure, but
  not granular enough
• Tools
• Models can contain all of the metadata required

8
What is a Model

• A model contains data defining the
  characteristics of a system.  This data is used
  as a representation of that system for the
  purposes of
• conceptual understanding of a system
• controlling the exchange of information with that
  system
• controlling the presentation of that system
  information to end users
• The 'data' is typically called 'metadata' in this
  context
• The information in a model must be complete
  enough to be useful for integration
• We need a way to understand relationships between
  models of disparate systems

9
Meta Object Facility (MOF)

• OMG standard used to define metamodels
• Metamodels define language/constructs to build
  models
• Relational for information sources
• BPEL, BPMI for business process
• XML Schema for XML documents
• UML for modeling applications
• Anything can be modeled
• MOF Metamodels are defined in terms of a common
  set of constructs (Package, Classes, Attributes,
  Associations, References, etc.)
• All MOF metamodels can be related

10
(No Transcript)
11
MetaMatrix Overview

• Leveraging MOF, MetaMatrix uses metadata and a
  model-driven approach to integrate and simplify
  access to disparate data.
• Model Driven Information Integration
• Federated Data Management

12
Model Driven Information Integration
A virtual representation of all physical data
sources - applications are decoupled from data
sources - details of data storage and retrieval
are abstracted - easily extended to new
information sources
13
MetaMatrix MetaData Modeler?
Platform-Independent Modeling

• Shows structure with class and package diagrams
• Displays in a consistent way metadata from
  different platforms
• Uses and retains domain-specific terminology

14
MetaMatrix MetaData Modeler?
Platform-Independent Modeling

• Shows structural transformations from one or more
  other classifiers
• Defines transformations with
• Selects
• Joins
• Criteria
• Functions
• Unions
• User Defined

15
XML Schema Mapping
Mapping exercise is no longer a black hole
16
How Model Driven Integration Works
17
Federal Enterprise Architecture (FEA) Reference
Models
18
The FEA MetaMatrix
Model Driven Integration
Agencies (Data Sources)
Lines of Business
Mappings
19
Key Points
20
An Enterprise Data Management Integration
Challenge ExampleThe New Department of Homeland
Security
21
MetaMatrix Integration Approach
22
The Model Driven Information Integration Process
Lines of business need access to integrated data
Separate agencies with disparate data sources
23
The Result Inter-Agency Information Federation
Emergency Preparedness Response
Border Transportation Security
Science Technology
Information Analysis
Virtual Views
Physical Sources
CIA
INS
Customs
FBI
Coast Guard
Secret Service
TSA
FEMA
National Guard
This approach is equally valid for intra-agency
data integration
24
Federated Data Management

• Web Services Approach
• Define XML Schema
• Map Schema to physical data stores
• Define WSDL
• Get XML document
• Standards based
• Seems simple, straightforward
• Mapping exercise can be a black hole
• But isnt if you manage metadata properly

25
Does it Work?
One of the world's leading financial management
and advisory companies, with offices in 36
countries and total client assets of
approximately 1.3 trillion.
The world's largest inter-enterprise software
company

• An agency with perhaps the worlds most
  daunting data management challenge and most
  demanding security requirements

26
Questions?
Ed Fakler ltefakler_at_metamatrix.comgt 301-208-6622