Internet Measurement: Infrastructure, Traffic & Applications, by Mark Crovella, Balachander Krishnamurthy, ISBN 0-470-01461-X, Wiley, 2006.
This book is a comprehensive reference guide to about 900 journal, conference, and workshop papers, and RFCs on the important and rapidly advancing field of Internet measurement. Interest in this growing field arises for three major reasons: commercial, social, and technical. Readers need nothing more than a keen interest in a methodical study of the subject matter from either a practical or research perspective to glean something from this book.
The book is centered on three architectural pillars relevant to measurement: infrastructure, traffic, and applications. Within each of these pillars, the topics are organized into four sections: properties, challenges, tools, and state-of-the-art. In the properties section, the authors review metrics that are important to measure in each area. In the challenges section, they discuss various difficulties and limitations that arise when trying to measure the metrics. The tools section covers some of the popular methods and products used to measure these metrics and work around the challenges mentioned previously. The intent is not to provide "user guides" for these tools. The state-of-the-art section presents the latest measurement results about covered properties and metrics, noting that they are subject to relatively fast obsolescence because of the rapidly evolving Internet.
The first three chapters provide background material. The first chapter provides an obligatory introduction to the Internet architecture, including how the "end-to-end" principle has been used for nearly 20 years to guide many design decisions in the Internet. The second chapter provides the analytic background necessary to study the Internet and cast its measurements in quantitative terms. The third chapter examines the nuts and bolts of Internet measurement, addressing the practical topics to consider in designing and implementing them, including the role of time and its sources.
The second part of this book also consists of three chapters, which cover the three pillars in depth. The first chapter defines metrics of interest for measuring the Internet and describes some of the barriers to their measurement, in particular "middleboxes," Network Address Translators (NATs), firewalls, and proxies that deviate from the end-to- end architecture principle, may block User Datagram Protocol (UDP) or Internet Control Message Protocol (ICMP) packets, or hinder visibility to endpoint IP addresses. The authors next explore various tools and methods for active and passive measurement, estimation, and inference of these metrics.
Readers may wonder why two important metrics are left out—router reliability and high availability—where Open Shortest Path First (OSPF) and the Border Gateway Protocol (BGP) "Graceful Restart" would be of interest.
The next chapter focuses on traffic properties that are important to understand, measure, and model. The authors examine the challenges in capturing, processing, storing, and managing large volumes of packets and flows, as well as those related to their statistical characterization. Readers engaged in data modeling and performance analysis will benefit from this chapter. The last chapter in this part of the book examines some popular applications: The Domain Name System (DNS), Web, and Peer-to-Peer (P2P). The authors discuss the shifts in application mix from the 1980s, when FTP was dominant, to the 1990s, when the Hypertext Transfer Protocol (HTTP) became dominant, to today, when by most accounts P2P is the dominant Internet protocol. Next, there is a thorough coverage of the what (properties), why (justification), and how to (tools) facets of measurement of the three popular applications, as well as some coverage of online games and streaming media.
The third part of the book covers material that spans multiple areas. Its first chapter deals with anonymization of collected measurement data, which arises because of the need for data sharing, while preserving identity-related, personal-sensitive, or business-sensitive information for applications previously examined. The second chapter provides a short—but important—coverage of the key areas where Internet measurement has played a role in security enforcement. Various attack types and tools to combat them are discussed. The third chapter examines numerous low-level monitoring tools for high-speed traffic capture, as well as an insightful look at the software architecture of two toolsets, dss and Gigascope, reflecting the experience of one of the authors at AT&T Labs with them. It also reviews some large-scale measurement platforms at the Cooperative Association for Internet Data Analysis (CAIDA), the Réseaux IP Européens (RIPE) community, and the High Energy Physics (HEP) community. The book concludes with a recap of trends, concerns, and emerging questions in Internet measurement.
The authors have blended their academic research and practical experience in Internet measurement and traffic modeling to provide the reader with a structured view to these vast subjects. I would have liked to see a more extensive coverage of Voice over IP (VoIP) and its associated performance measurement protocols, RTP Control Protocol (RTCP), RTCP Extended Report (XR), and RAQMON, given the gradual but inevitable shift of voice traffic from the Public Switched Telephone Network (PSTN) to the Internet with Session Initiation Protocol (SIP) peering.
Most probably, this book had already been published when the Federal Communications Committee (FCC) issued an order in May 2006 for all VoIP service providers to demonstrate compliance with the Communications Assistance for Law Enforcement Act (CALEA) wiretapping requirement within a year. This directive represents a notable departure from data anonymization principles covered in the book.
Overall, I consider this book an excellent reference source for diverse research and practical articles published in the field of Internet measurement. I highly recommend it to network planners, engineers, and managers responsible for instrumentation, traffic modeling, or performance analysis.
—Reza Fardid, Covad Communications