Integrating EDMSs and DBMSs - High-Performance Web Databases: Design, Development, and Deployment - page 537

The final step in the data analysis activity is to construct the data model.
Constructing the data model involves quantitative analysis. Using the
structure from the relational map and the number of accesses identified in
the transactional analysis, one derives a new structure for the model. This
new structure may result in new entities that may reduce the number of en-
tities that need to be accessed for certain high-usage transactions. The first
data model generally proves to be inadequate. Data analysis is therefore an
iterative process. As one proceeds through the iterations, one learns more
about the data. The new information may indicate that decisions made ear-
lier in the process may not have been optimal and may need to be revisited.

The ultimate database design will not only depend on the results of the
data analysis activity but also on the choice of DBMS. Good design does
not just depend on knowledge of the specific data requirements of a par-
ticular application or the general information requirements of an organi-
zation. These are critical elements of the design, but almost as important
is a good understanding of the particular DBMS, its architecture, and its
design constraints.

The critical aspect to understand about data analysis for the purposes
of this discussion is the process of functional decomposition. Functional
decomposition is a process that is extremely important to data analysis. It
is the process by which reality or a body of knowledge is decomposed,
summarized, or reduced into its most fundamental, elementary compo-
nents. This decomposition is generally from the one perspective that is im-
portant to the particular application being considered. These elementary
components are the data items that then ultimately make up the database,
such as those shown in Exhibit 1.

DBMS

Key(s)

Data Key(s)

Data Data

Exhibit 36-1. Elementary components.