Database modeling and design: logical design. (Q2574750)

This book covers the fundamentals of conceptual and logical design for relational databases in an example-driven textbook style. (Physical design is addressed separately in a companion book.) As stated in the cover text, the book is intended ``to help you grow from a new database designer to an experienced designer developing industrial-sized systems.'' To this end, the ER model and the UML are presented as languages for conceptual database modeling, and conceptual modeling as well as view integration and entity clustering are introduced briefly. Afterwards, the transformation of conceptual diagrams into SQL and relational database normalization are explained. In addition, the book includes a chapter on business intelligence and another one on CASE tools. Finally, an appendix summarizes the basics of SQL. While the book may serve as a starting point for database design, I doubt that it is an adequate source for industrial-sized database design. First, the design process does not take capabilities of modern relational database systems into account such as object-relational features and XML support, which have been standardized since SQL:1999, or active database mechanisms. Hence, transformation rules to deal with generalization and aggregation are presented, but native database support in terms of user-defined types and typed tables is not mentioned at all. Similarly and surprisingly, the importance of XML is only mentioned in the chapter on CASE tools but neither addressed from a design nor a storage perspective. Moreover, although the importance of tooling support is emphasized, object-relational mapping frameworks are ignored. In addition, a large fraction of the chapter on business intelligence focuses on how to build an OLAP engine, a topic which is not relevant for the target audience as admitted by the authors on p. 167, while commercially successful OLAP tools are not discussed, and smaller fractions address specialized topics such as forecasting and text mining (without mentioning text querying at all, e.g., via XQuery). Finally, the more formal part on normalization contains several errors: Not every relational table satisfies 1NF as stated on p. 109 (e.g., a 9-digit part number where the first two digits encode the part's type -- the authors missed the opportunity to warn against such combined domains), tables in SQL may contain duplicate rows (in contrast to relations mentioned on p. 110, which are irrelevant in the given context), 1NF is included in the definition of 2NF but not in that of the higher normal forms, and the example concerning the closure of FDs on p. 124 lacks all FDs with more than two attributes.