Enterprise Data Modeling Practices - High-Performance Web Databases: Design, Development, and Deployment - page 162

Quality of Data Model Contents
The quality of the contents of a data model is often intangible. Typical
errors can be seen in existing data models — for example, an account num-
ber is the number of an account. Definitions such as this often occur when
old pools of data have been reengineered in a sloppy manner. Data admin-
istration groups should make sure that data models are of high quality, and
if not, examine the reasons why. Sometimes projects lack the budget to do
a proper job, or employees may lack the motivation.
Rate of Change
A low rate of change, combined with efficient software development and
systems integration, is an indicator for high-quality data modeling. If data
modeling describes the core of the business rather than listing simple orga-
nizational facts, the indicator will show a decrease in change.
Managers of Data administration should ask themselves:
• Does the Data administration monitor the change rate? If so, is it an auto-
matic function?
• Is the change rate seen as a problem? If so, why? What are the reasons
for frequent change?
• Is the Data administration department clogged with change requests?
Is the Data Model Up to Date?
A data model should reflect the current state of the business for which
it is designed. The terminology must reflect the corporate facts and the
meaning of certain terms. The Data administration department should
ensure that the data model is fully up to date and carefully monitor the time
it takes to process a change request — the shorter the better.
Quality of Upper Model Levels
The upper level (i.e., A and B levels) of a data model should be under-
stood and approved by the users, because they are the people who run the
business and generate the profits that pay for Data administration. Data
administration should make every effort to find out what users think about
modeling efforts. It is also wise for Data administration to find out how
project teams perceive the quality of the data model.
Data Types as Entities
Data administration should know whether:
• There are data types that are being modeled as entities (e.g., records
that have no natural key, but a key that consists of all attributes).
• There are enumeration types coded as attributes that occur in several
entities.