Browse Books

Reviewer: John M. Artz

It is encouraging to see that the theory and practice of database design are taking another, much needed, step closer together. The unassuming cover and title of this book belie the fact that this is not just another textbook on database analysis and design. It is, instead, a compendium of the database designer's art, presented in a fashion that is as much needed as it is distant from the pragmatic, intuitive practice of database design today. In 578 pages, the author covers relational database design, semantic modeling, the translation from a logical data model to an implementation model, and implementation vehicles from B-trees to System R. This book puts together a lot of related and relevant information which has been previously treated as though it were from different domains. This could lead to a disjoint treatment of database design, yet the author supplies the necessary bridges and perspectives to connect the diverse theories. For example, he states: Semantic model abstractions are of a higher level order than data elements;:- 9T. . . Semantic models, however, do not usually possess criteria that can be used to prevent anomalies and eliminate redundancies. Since these criteria characterize relational normal forms, it is suggested that an ideal design process integrate the semantic model in the early design stage with the relational model in later stages. This is the mortar that holds the building blocks of database design theory together. The author clearly speaks to the database designer, addressing many difficult questions that the designer must face. Unfortunately, the text is probably too formal or too much like a textbook for the average practicing database designer. This introduces an interesting paradox. Anyone with the formal background to fully understand the text will probably be lacking in the development experience to appreciate the problems which it solves. Anyone who can appreciate the problems will probably be lacking in the formal training necessary to understand the solutions offered. This, however, is not a weakness of the text. It is probably more a comment on the distance between the theory of database design and the reality of the database designer in a large organization. While I would strongly recommend this book either as a text or simply for information, I must do so with some warnings. First of all, this is not an introductory text. There is an abundance of topics, all well presented, but covered in three to five pages when whole books have concentrated on introducing some of these topics. Many of the topics assume some familiarity. For example, the author emphasizes the synthetic approach to relational database design which relies heavily on the concept of functional dependency, rather than the more intuitive decomposition approach using normal forms which is more common in introductory books. The problems are ample and varied. Many require serious thought and there are no answers in the back. This, in conjunction with the concise presentation, might put a rather heavy burden on an instructor who is not well grounded in the full range of topics. I would also recommend this book for the practicing database designer despite its intensity. Organizations tend to pick up a methodology and run with it, or select an implementation technique and go off to solve every problem that doesn't run for cover. This book puts various approaches and techniques in perspective and increases the designer's chances of using the right approach at the right time. A lot of hard work must have gone into this book, and its shows]