Object-relational DBMSs: Tracking the Next Great Wave (Morgan Kaufmann Series in Data Management Systems) [Paperback]

Poor editing, however, abounds in this book. The reader must wade through paragraphs of opinionated comments, distracting (and often incorrect) pontifications and complaints. For example, hardware designers are roundly chastized for failing to support Multics-style protection-ring logic in today's microprocessors.

The text is heavily sprikled with superlatives such as 'explosive', 'incredible', 'astonishing', etc. It is irritating from a reader's perspective to be digging out good technical material betweeen snippets of pro-Informix marketing hype.

When not engaged in hype, the authors sometimes make curious and possibly false assertions. In one instance, they claim that Wal-Mart's huge data warehouse is utilized to 'rotate stock'. My uninformed suspicion is that they use it for much more than that.

Be prepared to read strange statements such as: "IBM is a hardware vendor - but they also happen to offer a relational database product".

Despite these destractingly spurious comments, I would recommend this text as good reading for those wishing to know more about object relational DBMS architectures.

Dr. Stonebraker is a pioneer in the field of object-relational database management systems (ORDBMS). He writes with confidence and clarity; he knows the subject matter because he was one of the early innovators. The book is mercifully short and to-the-point. It is well organized, and it includes an index, references, short code samples, and plenty of diagrams.

The book assumes you have some background and experience with existing relational database management systems and SQL. It would also help if you have some basic understanding of implementation details like B-tree indexes and cost-based query optimizers. You should also have an understanding of OO concepts like inheritance and polymorphism, and some experience with an OO language like C++, Eiffel, Java, or Smalltalk.

Armed with this background, the book is an excellent overview of the motivation, architecture, design, and features of ORDBMS. I especially liked the chapters on SQL parsing and query optimization, and also the discussions of how inheritance affects query and trigger processing.

If you've never been exposed to the ideas and concepts of "object-relational," this is an excellent starting point. If you have the background described above, you will be able to read this book in a few sittings and come away with a basic - albeit incomplete - understanding of the field of ORDBMS.

I say "incomplete," because Dr. Stonebraker fails to write about the dark side of ORDBMS. For example, once pointer-like references are introduced, the system runs the risk of 'dangling pointers'- precisely the kind thing a purely relational system elegantly avoids. This issue is never once mentioned, and yet I know for a fact it is a danger that must be dealt with on one commercial implementation I have used. There are other issues to be confronted and understood. C.J. Date has written extensively on the subject. Although C.J. Date's writing is sometimes long-winded and pedantic, you won't truly understand the ORDBMS topic until you've read what he has to say.

The book is also a compelling marketing promotion for the product once sold by Stonebraker's now-defunct company. Given the "feature matrices" and other comparisons in the text, it is clear that his product was the hands-down winner at the time the book was written. This is all ancient history, however. The book shows its age and bias. First of all, Stonebraker's company Illustra (later renamed Informix) was recently acquired by IBM. Secondly, very little is said about Oracle, which has since implemented a very respectable OR system of its own. Lastly, most of the other vendors he mentioned have fallen completely off the radar map. These days, all the database vendors have most of their attention turned towards XML and raw OLTP throughput. So it seems like "The Next Great Wave" predicted by Stonebraker never fully materialized. Perhaps the 2nd edition of the book has updated the vendor list and features matrices.

I wish more was said about his early work on the "POSTGRES" system. If you're interested, you can find out much more about POSTGRES and its open-source offspring, PostgresSql, by using any Internet query engine. There is now a whole cottage publishing industry built up around PostgresSql. If you want to experiment and explore ORDBMS, PostgresSql may be a good choice.

Based on what I can see in the table of contents, the 2nd edition doesn't include too much new material. Dr. Stonebraker has added a chapter on application servers. I know from his other work that he argues that middle-tier application servers are wrong-headed, because business logic should reside in the database itself - where it is physically close to the data.

I rate the 1st edition of the book 3 stars because (1) It is dated (2) No space is allotted to the problematic issues of ORDBMS (3) "The Next Great Wave" never seemed to take off as predicted.

Nevertheless, it is a worthwhile and useful book to read if you are interested in the topic. The 2nd edition will surely have an up-to-date comparison of the ORDBMS vendors as of 1999.