Databasics clearly explains the key concepts users and database professionals need to understand in order to build well-designed databases that answer business questions accurately and efficiently. Fabian Pascal, one of the industry's leading experts, identifies ten critical, recurring issues that both database users and vendors often fail to address appropriately. Pascal demonstrates why understanding these fundamentals is so important, providing detailed examples and solutions designed to help users escape the key pitfalls of database development. Among the topics covered: unstructured data and complex data types; business rules and enforcing data integrity; keys; duplicates; normalization; entity subtypes and supertypes; data hierarchies and recursive queries; redundancy; quota queries; and how to handle missing information. Along the way, Pascal offers no-holds-barred assessments of how well current SQL implementations and commercial products address each issue. Databasics, in...
Fabian Pascal is a highly regarded writer, well known as a staunch defender of the relational model's virtue. His standing in the RDBMS community is illustrated by C J Date's forward. If E F Codd is the father of the relational model, Date is certainly its uncle. Pascal seeks to clarify areas of database implementation that frequently prove problematic. This is not a beginner's book but is aimed at the experienced reader wanting to understand more about the underlying relational model and to press that knowledge into service.
The issues range from normalisation, keys, duplicate rows and missing information to business rules and integrity enforcement, "unstructured" data and complex data types and quota queries ("what are the top ten best selling products?", for instance). Each chapter covers one issue and comprises an overview, how the issue is best addressed, a demonstration of the practical benefits, the pitfalls of not addressing it and any pertinent recommendations. It makes for interesting reading. The layout and style, however, can irritate. Pascal uses many quotations but persistently [wedges] in his [own] wording: "if the DBMS knows the [integrity constraints] for the input [base tables] and relational operators [that define the view]." He also gives each chapter its own bibliography, which is only a problem because he has also chosen to cite separately each article from the wonderful Relational Database Writings series by Date et al. It all begins to look horribly like padding and Pascal's undoubted erudition doesn't need it. By this reviewer's count, he makes 47 references to the four books in the series and 11 to his own 1993 publication; furthermore, the ludicrous situation arises where two citations, identical in all but footnote number, appear consecutively. Whatever happened to ibid.?
Despite these misgivings, this book is stimulating and informative for anyone in search of further enlightenment on the relational model. --Mark Whitehorn