This book addresses the subject of performance analysis of computing systems and networks. The method of presentation is slanted heavily toward working out solutions to sample problem; for example, analyzing a three-stage pipeline of processors, or a transaction processing system involving a series of clerks, processors, and disk storage. The emphasis is on doing performance analysis, rather than laboring through lengthy mathematical derivations. Nevertheless, a limited number of derivations of key results are contained in the book, but derivations of more complex nature are not provided.
The material in the book unfolds from the simple to the more complex. The book is organized in three major parts. The first part, including Chapters 1 through 5, addresses the general concepts of computer resource allocation policies; basic performance measures such as system utilization, response time, and throughput; concurrency and parallelism in systems; data analysis and simulation; mean value analysis; and office communications. The second part, extending over three chapters, addresses Jackson network analysis and applications. The third part, in the last three chapters of the book, presents examples of priority scheduling of a single serially-reusable resource and introduces basic concepts in local area networks performance analysis.
Some of the examples that are worked out in the book are excellent portrayals of everyday problems encountered by designers of computing systems and networks; such examples may be used as a guide for solving actual problems. For instance, the section on Local Area Networks (LANs) unfolds a step-by-step analysis that evaluates the pros and cons of a small business system vs. a proposed LAN approach. Unexpectedly, the conclusion from the analysis demonstrates clearly that the disk storage device in the example is the bottleneck and a LAN will not significantly improve performance.
The book may be used as a textbook in a formalized program of study, or as a reference material for review by the practitioner. The problems following each chapter are interesting and comprehensive. In order to derive the most benefit from this book, the reader must possess a good background in analytical methods, particularly probability theory and queueing analysis, or else the reader is likely to get lost. Although the book is labeled “A Primer,” some of the material appears too complex for a primer. Also, some examples are not adequately explained and might be difficult to understand for the uninitiated reader. This problem evidently was recognized by the authors, who made sure that many references to additional reading sources were provided in appropriate points throughout the text.
Nevertheless, despite the rather technical nature of the discussions, this reviewer found the text pleasant reading owing to the clear and freeflowing style and occasional sprinklings of humor; for example, “How do we interpret data? With great care. . . .”
Some areas may be improved in future editions. First and foremost, the book is very light on measurement tools. A book of this nature should have more adequate coverage on measurement tools. One short paragraph covers hardware monitors, and there is one similarly short paragraph on software monitors. Nothing is provided on tools to measure network performance. In addition, there is a very light treatment of simulation. This reader felt that a book of this nature should have more on the latest in simulation tools, techniques, and applications. Similarly, the discussion on LAN, covered in the last chapter, is too cursory and appears to have been added hastily at the last minute as an afterthought.
Furthermore, several areas are either totally missing or very lightly touched on in the book. These areas include software analysis, network control programs, value added networks, reliability/maintainability/availability analysis, performance analysis standards, and distributed systems. It is hoped that these areas will be covered in future editions of the book.
It is also hoped the future editions will take care of several serious technical errors--errors which might be especially misleading to the novice reader. Several errors exist in a simple GPSS program appearing on pp. 121 and 122. Incidentally, aside from the errors, this program is of no value to a reader who does not know the basics of simulation and simulation languages. More explanation should be provided. Errors also appear in two FORTRAN programs in the text. Here, too, the value of having these programs in the text without any explanation is highly questionable; it is not clear how the programs relate to the formulas in the text; the treatment in this area is rather misleading and confusing. The book also contains some typographical errors; see for example, the third line on p. 93; the seventh line on p. 105; and the fifth line from the bottom on p. 385.
This reader evaluated the book primarily from the point of view of its usefulness to the practicing engineer. In conclusion, the book provides an overall review of techniques and methods in analyzing computer communications networks. Some of the examples in the book are excellent indeed; however, the book cannot be used as a comprehensive handbook for solving actual everyday problems. It is, rather, a collection of some analytical methods (e.g., probability, queueing theory) as applied to performance analysis of computer networks.
