Enterprise Data Modeling Practices - High-Performance Web Databases: Design, Development, and Deployment - page 160

of many corporations. To adapt a reference model to individual needs, it
should not be made more abstract, but it should be simplified to adapt it to
the needs of the organization that uses it.

Rebuttal: Reference data models, also
called application platforms, are often mistaken for off-the-shelf, prefabri-
cated data models intended for use in every financial institution. However,
they are very useful, and can be tailored to specific organizations.

Reference Data Models Are Useless.

The question most of the critics leave
unanswered is, what happens if data modeling is not performed? The first
alternative is to build island solutions, as in the pioneering days of data
processing. These islands are connected via interfaces. The lack of com-
mon terminology leads to problems and considerable effort. The negative
experiences with these solutions resulted in a data modeling approach to
fix those problems.

Alternatives to Data Modeling.

Another approach is to take business processes as the fixed point of
analysis. Processes can be analyzed, and in most cases, elementary pro-
cesses will be found that will be grouped in entities. This approach is a
purely dogmatic, bottom-up, object-oriented approach.

A practical approach is somewhere between bottom-up and top-down.
A healthy mixture of a top-down approach, manifested in level A and B
enterprise object models and bottom-up methods represented by project-
driven level C logical models, should be much better than any pure,
dogmatic approach.

ORGANIZATIONAL ASPECTS-DEFINING GOALS

A formal definition of goals and objectives for data modeling may seem
bureaucratic. But a short statement defining the goals of data modeling in a
company can be helpful for new employees in the Data Administration depart-
ment or for the more experienced employees as a reminder of their goals.

When looking at an organization's data modeling procedures, quality and
technical issues pertaining to the integration of data modeling should be
checked in the software development organization. The following list of ques-
tions is derived from the criticisms and rebuttals of data modeling and should
help organizations identify weak spots in their data modeling practice:

• Is there a written definition of the objectives for the Data Administra-
tion group?
• Are the goals consistent with goals stated in this chapter?
• What are the project goals of systems development projects in aspects
of data modeling?
• Is there a top-level data model that is publicized and available?