Carrier-Scale IP Networks: Designing and Operating Internet Networks , edited by Peter Willis, ISBN 0-85296-982-1, The
Institute of Electrical Engineers, London, United Kingdom, 2001
My heart jumped when I saw the nondescript brown box, about the thickness of a book, sitting by the receptionist. It was finally
here! I had waited almost two months in great anticipation for this book to show up. Was it going to be the all-encompassing
handbook for the network designers, operators, and managers in large-scale IP environments? The first few lines in the text
indicated that it just might be: "The aim of this book is to give the reader an understanding of all the aspects of designing,
building and operating a large global IP network."
The definition of "large-scale" as given by the author and for the purposes of this review follows: Provides services for millions
of end users, high-speed (greater than 100 Mbps) transit services, and is reliable, scalable, and manageable.
One thing to keep in mind is the way this book was constructed. The 16 chapters had 29 authors. Almost all authors came from some
area of British Telecom (BT) and all were subject matter experts in the chapter they wrote. The 16 chapters are grouped roughly
into four sections: Designing and building IP networks, transmission and access networks, operations, and development of future
networks. Sadly, all of this is squeezed into 293 pages.
For the reader new to designing and building large-scale IP networks, the first few chapters are gold. For the reader already
experienced in this area, it may bring back nostalgic feelings for the good old days of exponential growth. A lot of ground is
covered, including the obligatory overview of IP, sufficient enough to give a nontechnical person the key concepts of IP routing,
but can by skipped by those with even basic knowledge in this area. The examples given throughout this chapter (and the rest of the
book) come directly from the design of BT's and Concer's backbone. A whole chapter, "The Art of Peering," not to be mistaken for an
excellent paper of the same name [1], gives excellent key concepts in peering. Some coverage is even given to the logistics and
difficulties in building points of presence globally, going so far as to mention earthquake bracing for equipment bays.
The next set of chapters give the reader detail about the transmission network (for some, be prepared to think Synchronous
Optical Network [SONET] when you read Synchronous Digital Hierarchy [SDH]), and access networks, including various
forms of broadband, wireless, dial, and satellite.
The technical information was squeezed into these chapters, not enough for a good technical treatise, but enough to give readers
good grounding in a technology that is unfamiliar to them. The coverage was closer to being marketing material. These chapters
alone are not enough to bring those new to the field up to speed if they are to design or operate such a network.
BT opened itself up and gave us a view into the operations of its network. Individuals who have worked in an environment like this
will find something familiar. We get to see how BT structures the people, processes, and technologies. This is something that is
not usually open to inspection by people outside of an organization. Planning and developing the operations side of the house is a
difficult job. These chapters may give a kick-start to those coming into such a role.
I was disappointed with the two final chapters. Of course anything listed as being "the future" will one day become the present,
but I digress. These two chapters seem like the odd couple that just did not fit with the rest of the chapters. The first chapter
is on Traffic Engineering. It is really a primer on Multiprotocol Label Switching Traffic Engineering (MPLS TE). The
second chapter covers Virtual Private Networks (VPNs), both the MPLS and IP Security (IPSec) types.
The authors set out with a lofty goal, and did not quite hit the mark. This book would be appropriate for someone trying to get a
feel for what goes on inside of a carrier-scale network. People already in the business would be better served by just paying
attention to what goes on around them.
Perhaps a small focused group could set out to create a book (or should I say tome) covering the elements of design, the foundation
of support, and the basics of management. Something timeless is required here, independent of the protocol du jour, to develop the
next generation of competent netheads.
[1] "The Art of Peering: The Peering Playbook," William B. Norton, Equinix