The New Books and Multimedia column contains brief reviews of new books in the computer communications field. Each review includes a highly abstracted description of the contents, relying on the publisher's descriptive materials, minus advertising superlatives, and checked for accuracy against a copy of the book. The reviews also comment on the structure and the target audience of each book. Publishers wishing to have their books listed in this manner should send copies and appropriate advertising materials to Ioanis Nikolaidis at the address below, with an indication that books are intended for the IEEE Network New Books and Multimedia column. Appropriate books will be reviewed in the column.
Ioanis Nikolaidis
Computing Science Department, University of Alberta,
Edmonton, Alberta, Canada T6G 2E8
Infiniband Network
Architecture
MindShare, Inc. Tom Shanley, Addison-Wesley, 2003, ISBN 0-321-11765-4, 1208 pages, softcover
Infiniband Network Architecture (INA) is the latest book in MindShare's Inc. series of books on rapidly evolving technologies that are geared toward the commodity computing market. However, in contrast to interface architectures such as PCMCIA and PCI, INA is a perfect example of a fairly elaborate protocol stack designed almost from scratch with certain performance requirements in mind and the intention to support a wide variety of physical media, from copper to fiber, and in bit rates that range from 2.5 to 30 Gb/s. Hence, even though INA is geared for interconnection between processing units and I/O subsystems, it illustrates that protocol design is alive and well. One realizes that outside the IP-centric world, there is a world of special-purpose protocol standards that are both interesting as case studies of protocol design and commonplace, or at least becoming commonplace. Other than this observation, the book is definitely tailored, as the rest of the series, to the system architect and software developer (at the device level) who wish to support INA. Roughly, the topics covered are packet addressing, channel adapters, the role of switches, routers, and repeaters, the queue pair concept (its creation and operations), transfer types (including datagram, connected, and raw datagram service), send/receive operations (e.g., send, RDMA, read and write, atomic read/modify/write), link and physical layer descriptions, and subnet local and global addressing.
Network Management, MIBs and MPLS
Stephen B. Morris, 2003, Prentice Hall, ISBN 0-13-101113-8, 379 pages, hardcover
Network management is often mistakenly equated to a particular protocol, Simple Network Management Protocol (SNMP). The truth of the matter is that SNMP provides a platform based on which network management applications can be developed; as such, SNMP alone is still a distance away from management applications that can deal sensibly with the management tasks encountered in large networks. Stephen Morris is interested in the management of large enterprise and service provider networks, and he details in his book how to write such applications. There is a tendency in the book to avoid technology-specific aspects (e.g., specific link technologies) and the antithetical need for a certain technology that forms a common thread across all examples and a variety of network elements. This is the reason multiprotocol label switching (MPLS) is employed (i.e., as the common thread), especially in the examples provided. MPLS also helps abstract away from the differences across multiple link technologies, as they tend to coexist in many of today's multiservice devices. The main audience is software developers and network managers, but ample introductory material makes the book accessible to the first-time developer of network management software. Specifically, the first two chapters are an overview of SNMP and management information bases (MIBs), and also help identify the differences between applications and the special group of network management applications. The rest of the chapters cover guidelines and strategies for solving network management problems, the scalability issues, and the skill sets necessary to develop such software. The last four chapters describe the software components that participate in an NMS, and certain basic applications using C++ and Java, with code examples provided in appendices and online.