The publication of this book, covering DB2 Version 5 for MVS, IBM’s flagship enterprise (mainframe) database, is a welcome event. It is a significant sequel to Wiorkowski and Kull’s earlier book [1]. The latter was the text that launched my hands-on experience with DB2 nearly ten years ago, and I was interested to see whether the sequel lived up to the high standards of the previous work. It does.

The introductory chapter, “Concepts and Components,” covers important background and details in significant depth, namely the catalog, directory, SQL, bind, optimization, RID, and EDM pools. New or newly updated features include stored procedures and the subsystem architecture required to support them.

The chapters “Creating Tablespaces,” “The Basics of SQL Data Manipulation,” and “Index Design for Performance” cover essential material of interest to both beginning and advanced students. The chapter on index design covers all the essentials, including Type 2 indexes, synonym chains, and illustrations with attractive grayscale charts and graphs. The chapter on creating tables and views contains important aspects of relational theory--domains, referential integrity, and treatment of nulls.

“Creating Storage Groups, Databases, and Buffer Pools” is one of the more technical chapters. This material contains useful explanations of the storage hierarchy--main memory, extended memory, external disk, tape, and optical CD--and shows how to apply each to advantage. If you were wondering about the difference between virtual pools and hiperpools, read this material.

Stored procedures were introduced with DB2 Version 4 (MVS) as part of a phased implementation. But with Version 5, developers can now open a cursor (a position against a results set) and tab through the results set one record at a time. Instead of using a proprietary scripting code, DB2 uses C, Cobol, Assembler, and PL1 with Language Environment 370 as the scripting language. (Watch for Java to be added to this list.) This versatility is underappreciated as a powerful push in the direction of open standards. The exposition in this chapter is clear and illuminating.

“Batch Processing” contains a study of restart logic, which can be complex. Getting it right has major performance implications, since rerunning long processes in total is the very paradigm of bad performance. “Concurrency Control” provides a briefing on crucial performance tradeoffs. For example, do you think that row-level locking always increases concurrency? Think again. It might have its uses when everyone is trying to update the same page, but that rarely happens even with medium-sized tables. This was apparently one of those enhancements requested to eliminate perceived advantages of other database engines.

“Parallel Processing” covers I/O parallelism, CPU parallelism, and Sysplex parallelism (in which multiple SMP engines are coupled with shared global memory). The requirements for CPU parallelism, especially Type 2 indexes and what that means, are covered.

The optimizer is one of the most important components for producing performance in DB2, and the chapter on it is likewise one of the most important. This insightful and detailed chapter explains the access path chosen by the optimizer, a component rightly regarded as one of the innovations of DB2 and a core competency of IBM’s DB2 labs.

One of the strengths of the earlier book was the SQL statements--in effect templates--provided to query the relational catalog. Frankly, this SQL gets long and complicated, which is where an electronic version from which one could cut and paste would come in handy. Fortunately, one is available, since the book now can be obtained as a CD-ROM or as downloadable content from a Web site (www.Gabrielledb2.com). Quarterly updates based on the latest releases of DB2 will be available.

“Copy, Quiesce, Report, and Recover” contains one of the best short comparisons of concurrent copy and image copy available in the literature. This chapter contains much useful introductory material on the utilities runstats and reorganization, as well as important advanced material on online reorg with concurrent update and inserts.

Excellent charts and examples showing performance results and including simple tweaking of DB2 that can provide integral factor improvements in response time, throughput, and availability can be found throughout the text. For database administrators, developers, and more technical managers, this book is the definitive look under the hood into the mechanism that drives the operational engines of DB2. Those responsible for squeezing performance out of DB2 should get access to it. Study it, and reap the benefits.