Designing Relational Databases - High-Performance Web Databases: Design, Development, and Deployment - page 406

• Developing production databases. The designer will need all steps of a re-

lational database design methodology, with additional product-specific
techniques.

• A designer can eval-

uate a DBMS by considering how effectively each step of a design
methodology can be performed using that DBMS for a given set of user
requirements. Compare multiple DBMSs by determining which prod-
ucts require the most work (e.g., the most customized code). This
comparison will indicate which products enable maximum satisfac-
tion of design requirements and with what degree of ease.

Assessing the maturity of relational DBMS products.

• Evaluating the applicability of automated design tools. A designer can

evaluate a design aid by analyzing the degree to which it automates
the steps and guidelines of a relational database design methodology.
The methodology should provide an effective benchmark of the prod-
uct's ability to meet design needs.

• Establishing a basis for database design reviews. When a standard

step-by-step design process is used, database design reviews are
more straightforward because all participants are grounded in com-
mon design approaches and goals.

SUMMARY

Relational technology can be an important asset to an organization,
especially if it is used to build relational databases that serve a variety of
business needs. To be used most effectively, relational technology requires
a data-driven design approach.

This chapter describes data-driven relational database design as com-
posed of two phases. The first, translation of the data model, preserves the
logical structure and integrity of the data. The second phase, tuning,
refines a preliminary relational design to accommodate performance and
functional objectives.

A relational database design methodology should consist of step-by-step
procedures and supporting rules. Use of these steps and guidelines will
increase the productivity of relational database designers. In addition, it
may assist them in avoiding disasters and less severe (albeit more com-
mon) errors. A methodology should encourage use of consistent and
proven approaches for resolving technical issues. It also should suggest
practical techniques for meeting functional and performance challenges
posed to the relational database products of today and tomorrow.