2. Abstract
• Data is a company’s greatest asset. Enterprises that can harness
the power of their data will be strategically positioned for the next
business evolution. But too often businesses get bogged down in
defining a data management process, awaiting some “silver bullet”,
while the scope of their task grows larger and their data quality
erodes. Regardless of your eventual data management solution is
implemented, there are processes that need to occur now to
facilitate that process. In this webinar we will discuss using your
current data modeling assets to build the foundations of strong
data quality.
PAGE 2
3. Biography
• Victor Rodrigues brings 10 years of experience of advanced usage
of the CA ERwin Modeling suite first as a Senior Support Engineer
for the CA ERwin Modeling suite of products and currently as a
Senior Software Engineer for Programmer’s Paradise. In this time
he has used his extensive experience to implement the tool with
various large and small enterprises. This experience includes
customization of the CA ERwin tool via the API and Forward
Engineering template editor as well as maximizing modeling by
integrating the product suite which includes CA Model Validator,
CA Model Manager, CA Process Modeler, SAPhir, and now CA Data
Profiler.
PAGE 3
4. Agenda: The Road to Data Quality
• Start Trusting Your Data
• Obstacles & Object Lessons
• Essentials
• The Data Quality Steps
PAGE 4
6. Data Quality Realities
• Data is a company’s greatest asset.
• Accenture survey shows 40% trust “gut” over BI.
• 61% say good data was not available.
• Data plus quality equals information.
PAGE 6
8. Obstacles to Data Quality
• People, Process or Software related…
– All of the above.
PAGE 8
9. Silver Bullets?
• Isn’t the Data Warehouse/ERP solution supposed
to be doing this?
– Definitions can be context specific.
– Delays taking ownership of your data.
Nike/I2 CMS example.
PAGE 9
11. Data Governance Essentials
1. Metadata Standards
2. Collaboration
3. Structure
4. Policies and Standards
5. Cultural Change
6. Getting from “as is” to “to be”
PAGE 11
12. Data Modeling as the Hub
Application Development
Business Intelligence (BI)
ERP
Data
Model
Database Management & Data Warehouse
Administration
Master Data Management (MDM)
PAGE 12
15. Why Metadata Matters
• Start by Defining Meta Data
– Disagreements as to what a definition is
• Too Conceptual – Definitions are not possible
• Too strict
– Everything can be defined.
PAGE 15
16. Strict Yet Flexible
• Too Strict Example.
– Phone number as a single entry.
• Too Flexible.
– Phone number as XML?
PAGE 16
22. 2 - Collaboration
• Share designs and templates.
• Model lineage and history.
• Centralized reporting.
PAGE 22
23. Overcoming Silo Mentality
• Director of National Intelligence
• “A Space” encourages collaboration.
PAGE 23
24. Collaboration
• Updates to apps migrate to source DBMS models
and vice-versa.
• Define and enforce your glossary and standard
abbreviations.
• Create templates.
PAGE 24
25. 3 - Organization
• Build on Existing Processes
– You are already governing data (informally).
– Identify your assets.
PAGE 25
26. We Need Structure
• Add structure to your existing process.
• Link your models.
• Create libraries in your Model Manager that
contain linked application models, related DBMS
models, etc.
• Create your Model Manager security roles.
PAGE 26
29. 4 - Enforcing Standards
• Generate diagram and repository reports to other
teams.
• Promote your value to your Business Analysis
teams.
• A bidirectional hub to report your standards and
update your policies.
PAGE 29
30. 5 - The Hard Part – Cultural Change
• Data Quality requires a change of culture.
• There is no silver bullet. It is a process.
• Like any habit, it becomes easier with time.
• Replacing bad habits with good ones.
• The process must me bottom up and top down.
• NUMMI plant example
PAGE 30
31. Good Habits
• Model Everything • Own your (meta)data.
– Applications – Be a good shepherd.
– DBMS – Do not pass along bad data.
– Data Warehouses
– ERP systems
– Others
• NoSQL databases, UML
models, etc.
• Model your Data Entry.
– Valid Values?
– Nullability?
– Proper and matching
PAGE 31
Datatypes/Domains.
32. 6 - Create Your “TO BE” Design
• Create the “To Be” model.
• Compare “As Is” and “To Be” environments
• Create a process.
PAGE 32
33. Conclusion
• Treat data like the asset that it is.
• Data quality creates information.
• Strong metadata definitions + good habits = data
quality.
• Data modeling allows us to structure our
metadata.
• Data quality is a process and requires cultural
changes.
PAGE 33