Metadata Strategy:

1
Metadata Strategy

• How to Pick an Approach
• Capital One - Jason Legum

2
Problem How do we know what we know?
3
Metadata Tells Us What We Know

• Business metadata
• Definitions of data elements
• Valid values and valid value definitions
• Technical metadata
• Physical structure layouts
• Data transformations

4
Challenge 1 Audience Definition

• Application challenge
• An IT solution?
• A Business solution?
• An IT solution for a business problem!

• Content challenge
• Who owns metadata?
• Who owns data?
• Who owns applications?

5
The Quality Imperative

• Low quality metadata spawns
• Replicated repositories
• Inconsistent metadata
• Competing systems
• Low successes
• High quality metadata encourages
• Confident development
• Teamwork across the enterprise
• Metadata beyond definitions

6
Challenge 2 Limits on Quality

• A perfectionist will never collect metadata
• Identify metadata quality
• Provide a feedback mechanism
• Collect, perfect, protect your content

7
A Solution is not just technology
Steward Group
Vendor
Partnership
Organizational Change
Ownership
Performance Improvements
Metadata Repository
Users
Awareness
Production Services
SDLC Integration
Content
Support
Process
8
Solution Examples

• Business metadata
• Data definition changes
• result from code changes
• driven by business requests
• implemented by a turnover process

• Technical metadata
• Database design changes
• are stored in a catalog
• that can certify metadata repository content.

9
Metadata Tells Us What We Know

• Technical metadata
• Physical structures
• Data transformations
• Business metadata
• Definitions
• Valid values
• Organizational metadata
• Metric metadata
• Operational metadata

10
Metadata Informs in Great Detail!
11
Challenge 3 Broad v. Deep

• Broad
• A little metadata for every production structure
• Deep
• Detailed metadata for critical production
  structures
• Decision points
• What groups are experiencing the pain?
• What processes can be leveraged for maintenance?
• What metadata is readily available?

12
Solution Sets

• Departmental solution
• Applications metadata with few interfaces
• Inter-departmental solution
• Database metadata
• Enterprise solution
• Integration of applications and databases

13
Enterprise Target Architecture
Data Modeling
Process Modeling
BI Tool 1
BI Tool 2
Stat Tool
MD Quality
Standard
Reporting
Reporting
Metadata Store
Metadata Store
Metadata Store
Metadata Store
Metadata Store
Flat
Flat
ODBC
Files
Connector
Files
Import
Export
Import
Export
Import
Export
Import
Export
Import
Export
Metadata Repository
Online
User
Access
Oracle
DB2
Teradata
SQL Server
Copybooks / Text
Excel Template
Source
Scripts
Batch Jobs
Design
Run
-
time
RDBMS Catalogs
Flat Files
User Documents
ETL Development
Bulk Load
14
Challenge 4 Local v. Enterprise

• Reusing a solution maximizes investment value
• Too much reuse creates a maintenance burden
• People dont need to store their own metadata
• But greater exposure is free leverage
• except for performance, security, support
• and other issues around scale
• Aphorisms
• Recognize their limitations

15
Metadata Roles

• Data Steward maintain content in repository
• Data Modeler provide context for definition
• Data Owner authorize maintenance
• Developer implement communicate change
• Administrator manage repository tool

16
Challenge 5 Organizational Costs

• Technology costs are significant
• Hardware, software, and administrative costs
• Configuration necessary for any package
• Organization costs are greater
• New roles for existing people
• New processes to manage metadata
• Hard costs and soft benefits

17
Conclusions Crawl, walk, run

• Best solution leverages existing resources
• Leverage available processes
• Leverage existing repositories
• Gain commitment for on-going support
• Organizational pain funds solutions
• Emphasize the long term nature of the problem
• Implement a supportable solution
• Scope breadth and depth correctly
• Scope user community correctly

18
Questions?