A number of good books on the performance aspects of computer communication networks are already on the market, ranging from Kleinrock’s now-classic two-volume Queueing systems [1,2] to Schwartz’s recent work [3]. These books deal with such performance issues as network design, protocol analysis, and control algorithm evaluation.

The emphasis of this new book on performance evaluation, as the author points out in his preface, is on the user-perceived end-to-end performance of a common-user network. Certainly, a book addressing these topics will be timely and should fill a gap in what is already available.

Looking at the subjects covered by the author, my assessment is that the book has only partially achieved its objective. As described in Section 2.2, performance parameters fall into four categories: delay, throughput, accuracy, and availability. But only chapter 7, devoted to a discussion of delays encountered by character-mode and block-mode terminals, can qualify as covering end-to-end performance estimation. Of the remaining three performance parameters, the accuracy of transmission links is covered not from an end-to-end perspective but as a parameter of a network component. Verma offers very little treatment of the throughput parameter and just introduces the availability parameter without any further discussion.

The major merit of the book comes from chapter 8, “Performance Objectives.” This chapter considers the basis for deriving performance objectives from an engineering standpoint, using an economic argument. I wish the author had devoted more space to this important topic. Only through an understanding of the process of setting realistic performance objectives can a user make intelligent decisions “prior to a long-term and potentially heavy financial or operational commitment” (from the preface).

I agree with the author that, if this book is to be used as a classroom text, it must be supplemented by other current references. The depth of treatment is rather uneven. In some places, the mathematical derivations are quite involved, while in others, the analyses are trivial.