Internet Performance Survival Guide: QoS Strategies for Multiservice Networks , by Geoff Huston, ISBN 0-471-37808-9, John Wiley & Sons, 2000.
Many readers of IPJ are familiar with the name Geoff Huston. He contributes articles frequently. I find his style to be very lucid and his writings to be very well structured and organized.
I have need at my job to begin implementation of Quality of Service (QoS) strategies to deal with an ever-increasing demand for Virtual Private Network (VPN) tunnels over shared media. So, when I came across the title of this book and saw who wrote it, I jumped at the opportunity to review it for IPJ.
This book is organized more like a textbook than a reference manual. If you are looking for a quick and dirty guide that simply lists all the tricks of the trade and gives examples of how to implement them on specific equipment, then this book is not for you. If, however, you are looking for a well-written text that will help you to understand the issues, the practices that address them, and the theory that underlies these practices, then this is an excellent book.
The book begins with a chapter that explains in detail the problems that administrators and engineers on heterogeneous, multiprotocol networks face today. There is a quick historical survey of the evolution of networking and how that has shaped the nature of the problem. In a very topical fashion, this introduction covers the basic techniques that can be used to implement QoS, but also explains the complexity involved with these techniques, their limitations, and why they are not widely deployed yet. The book continues from there, starting with a low-level view of the building blocks of the network and gradually building to higher- and higher-level topics.
The second chapter begins with some details about the performance features built into the Internet Protocol, and in particular IPv6. This chapter continues into TCP and covers all the well-known performance features that are built into it, and then moves on to routing, switching, and Multiprotocol Label Switching , or MPLS. MPLS is a unified approach to switching across large networks, and it has particular applications to QoS. This topic is one of the main reasons I sought for this book, and I am glad it was covered in such detail. The second chapter ends with a survey of the various transmission systems that are available today, and discusses in detail the performance characteristics and problems that are peculiar to each.
The third chapter is a well-organized exposition of the various types of performance-tuning techniques that are available. The author keeps the discussion at a reasonably abstract level, yet is not afraid to discuss the details of the application of these techniques to the specifics of the network when such details are important. In particular, the use of QoS techniques in conjunction with the Open Shortest Path First (OSPF) routing protocol is discussed.
The fourth chapter combines the building blocks of Chapter 2 and the techniques of Chapter 3 into an architectural view that spans the network. The author discusses the metrics that can be used to analyze network performance, the protocols that can be used to implement service strategies, the tradeoffs that are inherent in the problem, and the policy choices that need to be made in order to come up with a clear design. In particular, the Integrated Service and Differentiated Service models are discussed separately, and then the author shows how these can be combined into an end-to-end network design. As with Chapter 3, the author explains important specific cases such as the use of the Resource Reservation Protocol (RSVP) with ATM.
The fifth chapter moves on to explain how the architectures that have been described can be used to attack the various kinds of problems that exist on real networks. The emphasis is clearly on the end user of the system and how to measure the levels of service being provided and to bring into play the techniques already discussed to assure a consistent level of service. The organization of this chapter seemed less clear than that of the previous chapters, but that is perhaps due more to the nature of the complexity of the problems being discussed than to the author's limitations or inattention.
The sixth chapter provides little new material, per se, and is more of a perspective on the material already provided. However, it contributes highly to the content of the book in two important ways. First, it provides more of a top-down view of QoS to complement the material in the preceding four chapters, which present a mostly bottom-up view. Secondly, it acts as a natural bookend for the first chapter. The first chapter raises the issues and poses the questions. The middle of the book examines the protocols, techniques, and architectures in detail. The last chapter then attempts to answer the questions that were initially raised.
The author does an excellent job of presenting material that is complex, vast, and is still in the process of evolving in the field. He is very diligent about managing the level of detail, and is careful to first cover the material topically before diving into the details. The examples are appropriate and have been carefully chosen.
One of the features of the material that is most appreciated is the practical perspective that the author brings to his work. The theory never gets out of hand, and is always balanced by a real-life approach to problems that, unfortunately, can never be completely solved. And, the author's observations always seem in tune with the experiences of the reader.
The material is well organized, and readers will appreciate the effort expended on the textual conventions that help to organize and structure the material. The diagrams that accompany the text are clear and well-placed, and they contribute to the reader's comprehension.
A glossary in the back helps a reader who has not thoroughly read the preceding sections of the book. The index is also well done, and the reference material is copious and pertinent.
Overall, I would recommend this book to any professional who manages large, integrated networks, particularly those professionals who work for Internet Service Providers in an engineering capacity. I think this reflects the particular interests of the author, but that is as it should be.
—David P. Feldman, Tudor Investment Corporation
Would You Like to Review a Book for IPJ?
We receive numerous books on computer networking from all the major publishers. If you've got a specific book you are interested in reviewing, please contact us and we will make sure a copy is mailed to you. The book is yours to keep if you send us a review. We accept reviews of new titles, as well as some of the "networking classics." Contact us at email@example.com for more information.