THE MAGAZINE OF USENIX & SAGE June 2002 volume 27 • number 3

inside: BOOK REVIEWS

& The Advanced Computing Systems Association & The System Administrators Guild the bookworm

by Peter H. Salus have to confess that I read and com- mented on an early manuscript and Peter H. Salus is a member of the ACM, wrote the Foreword to McCarthy’s book. BOOKS REVIEWED IN THIS COLUMN the Early English Text I have no financial interest in it.) Society, and the Trol- lope Society, and is a life member of the Cups of Java American Oriental J2ME is the Java 2 Micro-Edition: it is IP ROUTING Society. He is Chief Knowledge Officer at designed for resource-limited devices RAVI MALHOTRA Matrix.net. He owns Sebastopol, CA: O'Reilly &Associates, 2002. neither a dog nor a like cell phones or pagers. It is really Pp. 219. ISBN 0-596-00275-0. cat. cleverly designed. Topley has done the [email protected] excellent job that I’ve come to expect THE PROCMAIL COMPANION from the Nutshell handbooks and refer- ARTIN C ARTHY I’ve got a heap of books I want to talk M M C ences. Edinburgh, Scotland: Addison-Wesley, 2002. about this month. And a poster, too, for Pp. 235. ISBN 0-201-73790-6. those of you with space on a wall J2EE is the Java 2 Enterprise Edition. J2ME IN A NUTSHELL between “User Friendly” and “Dilbert” This tutorial is full of real examples and KIM TOPLEY clippings. thoroughly useful pointers on just how Sebastopol, CA:O'Reilly & Associates, 2002. J2EE can be used in your enterprise. The Pp. 450. ISBN 0-596-00253-X. Getting Around CD contains three (!) J2EE tutorials, There are currently over 175 million J2SE and J2EE software development THE J2EE TUTORIAL host machines on the Internet (up from kits, a sample Java BluePrints, and the STEPHANIE BODOFF, ET AL. Boston, MA: Addison-Wesley, 2002. 213 in August 1981). Getting mail, VoIP, Forte for Java plug-in. Whew! streaming video, http[s] packets, ftp and Pp. 491 + CD-ROM. ISBN 0-201-79168-4. For over five years, Flanagan has occu- scp packets, AUDIO, and other stuff pied a prominent place on my bookcase. JAVA IN A NUTSHELL, 4TH ED. from box to box is not trivial. I have AVID LANAGAN The 4th edition of his “Desktop Quick D F relied on folks like Huitema and Perl- Sebastopol, CA: O'Reilly & Associates, 2002. Reference” has gained a lot of weight: man to elucidate the complexities of Pp. 969. ISBN 0-596-00283-1. just over doubling the 438 pages of the routing in the past. Malhotra’s small 1996 version. Hardly any flab, though. EMBEDDED LINUX (barely 200 pages) book will now live CRAIG HOLLABAUGH next to my desk. Though ostensibly A Pair of Penguins Boston, MA: Addison-Wesley, 2002. Pp. 419. about Cisco routing, Malhotra’s exposi- ISBN 0-672-32226-9. It’s no secret that I’m a Linux user and a tions of RIP, IGRP, EIGRP, RIP-2, OSPF, Linux enthusiast. Over the past two LINUX ADMINISTRATION HANDBOOK and BGP-4 are good enough for those years or so, more developers have turned EVI NEMETH, ET AL. working with Juniper, Nortel, etc., hard- to Linux to provide solutions for Upper Saddle River, NJ: Prentice Hall, 2002. ware. Definitely a must have! Pp. 889. ISBN 0-13-008466-2. embedded systems. Hollabaugh’s pres- Incidentally, if you’re interested at all entation is more than merely adequate, VMWARE in Internet history, Peacock (www. but I have to admit that I still prefer BRIAN WARD peacockmaps.com) has a really fine “First John Lombardo’s presentation (pub- San Francisco, CA: No Starch Press, 2002. Pp. 249. ISBN 1-886411-72-7. Maps of the Internet” poster for $29.95. lished by New Riders last year). Just right for covering the hole in the And here’s the book for every penguin- INTRODUCTION TO PROGRAMMING plaster you tossed that CPU through . . . IN EMACS LISP, 2ND ED user! Nemeth’s UNIX handbook has ROBERT J. CHASSELL /dev/null spun through several editions since Boston, MA: FSF, 2001. Pp. 292. 1989. Each one more impressive than I get a lot of email. Much of it offers me ISBN 1882114-43-4 the previous. Now here’s the Linux ver- $$$ or XXX or and opportunity to buy sion, written by Nemeth and her col- Viagra or enlarge my breast size. For sev- leagues, Garth Snyder and Trent Hein. It eral years, procmail has been my friend: covers Red Hat, SuSE and Debian. It is I’m now down to about 100 messages a well-written. If you use Linux or if day; some of them actually meaningful. someone on your site uses Linux, this Marty McCarthy has written a fine book book is indispensible. Thanks, Evi. on setting up and using procmail. (I Thanks, Garth. Thanks, Trent.

56 Vol. 27, No. 3 ;login: book reviews

Virtual Machines Channel, SCSI, and TCP/IP. The level of the things that need to happen for IP coverage is introductory, and the knowl- SANs to succeed.

A virtual machine enables the operator EVIEWS edgeable reader can skim through these R to pretend to use one OS while running No book is perfect. In the explanation of on top of another. VMWare does this for chapters. But network experts learning OOK

RAID (Figure 3-1 in the book), a “0 + 1” B a variety of Windows platforms, Linux about storage or storage gurus who need RAID is shown as a mirror of two to learn about networking will find one and FreeBSD. Ward’s presentation is striped arrays (RAID 0, then 1). Almost or more of these chapters worth careful good, though I have a minor problem: no one implements it this way; instead, reading. my notion of how to “get the most out they stripe together a set of mirrored of Windows” is to just run something Chapter 8 is the heart of the book, where drives (RAID 1, then 0). On page 174 the else. iSCSI (Internet SCSI), iFCP (Internet lower drawing should read “Tunnel Fibre Channel Protocol), and FCIP Mode Security Association” as opposed Lisping (Fibre Channel over Internet Protocol) to “Transport Mode.”To spend more It has been over a decade between edi- are discussed and contrasted. Clark’s than a couple of sentences on the book’s tions of Bob Chassell’s Emacs Lisp book. extensive Fibre Channel background flaws would be an injustice to the The new edition contains a really fine really shines here as he is able to exam- tremendous collection of knowledge tutorial as well as the new features ine how these new protocols interact contained within. This book is destined included in GNU Emacs v21. It’s defi- with existing Fibre Channel systems. You to stand for many years as the reference nitely worthwhile. Bob, another fine can spend weeks combing the standards on IP SANs. piece of work. (I’ve done it!) trying to learn what Clark clearly explains in a few simple pages. SOFTWARE FOR YOUR HEAD: CORE IP SANS: A GUIDE TO ISCSI, IFCP, PROTOCOLS FOR CREATING AND AND FCIP PROTOCOLS FOR STOR- It turns out that these protocols do not really stand on their own, and the next MAINTAINING SHARED VISION AGE AREA NETWORKS three chapters are devoted to iSNS JIM AND MICHELE MCCARTHY TOM CLARK (Internet Storage Name Server), secu- Boston: Addison-Wesley, 2001. Pp. 464. Boston: Addison-Wesley, 2002 ISBN: 0-201-60456-6 ISBN: 0-201-75277-8 rity, and QoS (Quality of Service). The Reviewed by Steve Johnson Reviewed by Steve Reames iSNS protocol provides naming and dis- [email protected] [email protected] covery services that are inherent to Fibre This is one of the more unusual books Storage Area Networks (SANs), and Channel, but require an additional on software engineering available today. storage networking in general, are server for IP-based storage protocols. Some of the ideas it proposes are bril- becoming increasingly important com- The chapter on security discusses the liant, some are weird, and some are ponents in corporate data centers. Tradi- issue from two perspectives. First, secu- both. tionally dominated by Fibre Channel, a rity in Fibre Channel SANs is discussed. recent plethora of IP-based standards This gives the reader the needed per- In an earlier book (Dynamics of Software are competing to either augment or spective for the second section, which Development), Jim McCarthy states that replace the entrenched standard. Clark covers the same issues for IP-based “Software is the process of turning ideas has done an excellent job of putting SANs. into bits.”He goes on to make a convinc- together the important aspects of these A short chapter on Infiniband gives ing case that the big problem in software new protocols, and comparing and con- some insights about how this interface engineering is not the individual pro- trasting them in a fair and even-handed may or may not become part of SANs in grammers’ abilities to turn their individ- way. If you have anything to do with the future. The chapter on SAN applica- ual ideas into bits but, rather, “aligning storage networks, you need to read this tions is thorough and detailed, and all the ideas in the various programmers’ book. probably provides the most extensive heads.”If the ideas are aligned, any problems one person may have realizing IP SANs is just over 280 pages long and collection of real-world scenarios yet these ideas are quickly caught and reme- packed with illustrations. It would be compiled. Even if you have a Fibre died. If the ideas are not aligned, prob- easy to jump into the technical details of Channel SAN and intend to stay with it, lems will be frequent, contentious, and the standards, but instead Clark takes his you need to read this chapter to under- difficult to find and eliminate. time and covers the background mate- stand how the industry is developing. A rial that will be needed later. The first final chapter of conclusions examines Software for Your Head is an attempt to few chapters cover shared storage, Fibre the potential of IP SANs, and outlines provide some algorithms or patterns a

June 2002 ;login: SOFTWARE FOR YOUR HEAD 57 book reviews

group can follow that will lead to this The authors also believe that when less obvious example can be found by alignment of ideas. The underlying someone is “out,”it is often because they going to an opera (turning ideas into premise is that most groups are working are feeling some emotion (work-related notes), where hundreds of people exe- at a fraction of their possible potential, or not) that is interfering with their full cute virtuoso feats with split-second and by improving the group functioning engagement. So one way to allow more accuracy over a span of several hours, you can get large productivity gains and people to be “in” more of the time is to with rarely any major errors. Most soft- increase everyone’s job satisfaction. Peo- allow emotions, even negative ones, to ware teams would literally think this is ple who have worked in both high- and be acknowledged within the group. They impossible. low-functioning groups will have no are not suggesting that software teams If, as I believe, cooperation is built into trouble with this concept. The key to the become encounter groups or therapy our nervous systems (like language and “software problem” is not reading groups – far from it. Just that it should the ability to reason), creating an effec- another book on multiple inheritance, be OK for someone whose daughter is in tive group should be as easy as creating a or even using the latest requirements the hospital to acknowledge it in the safe place for our “cooperation genes” to language, but getting the group aligned group, and through this honesty help express themselves. And, sometimes, it is and functioning at a high level. the group to become stronger. this easy. By this time, many readers are probably Another idea put forth in the book is Of course, musicians are trained from stifling yawns. There are lots of “team- that the “product” of management is, in early on to communicate with other work” workshops that teach people their fact, the group behavior. So if manage- musicians, the conductor, and the audi- Meyers-Briggs scores and purport to ment is unhappy with the group, it ence. Athletes in team sports get similar teach “trust” by hanging on ropes. The needs to look at its own actions and atti- training. Programmers, on the other authors are scornful of such remedies, tudes for the root cause of the problem. hand, are still taught as if success which typically do not hold individuals This is the kind of obvious statement depends solely on one’s ability to under- accountable for their actions in the that never seems to be obvious to the stand multiple inheritance, data struc- group and do not articulate the desired managers themselves. tures, and complexity theory. Not only group behavior clearly. The description of group maladaptive do students not learn to cooperate, but One concept in particular struck me. patterns, and how to break out of them, their professors – role models – function The authors believe that every group accounts for some of the most brilliant in universities that often prize individual functions at the level of the least enga- and insightful writing. The authors are achievement over cooperation as well, ged person. In many meetings, some frequently quite vivid in their language. reinforcing this isolationism. people want to be there and so have a For example, they discuss the phrase “He Is it surprising, then, that many software vested interest in seeing the issues gets an A for effort.”In other words, they people find working in a group boring, resolved; others are just warming chairs. point out, he failed, and he wasted a difficult, and frustrating? And their These disengaged people may see things great deal of time and energy failing. managers, promoted because they were that they could contribute but have the They prefer the motto “Fail Cheap.” the best coders, are not typically trained attitude, “Why should I speak up? It will I found the authors to be very effective to break this cycle. In fact, the good pro- just prolong the meeting . . .” in describing the problems that hold grammers with integrity often turn The authors see this attitude as a loss of groups back and articulating how an down or resign from management jobs, integrity. If people do not function at effective group would function. I have recognizing that they are contributing to the highest level, they are weakening more difficulty with the remedies they the problem rather than the solution. themselves and the group. And, more to suggest. These are overly structured, With this background, perhaps some the point, they hold the group (and even draconian, for my taste. They may draconian techniques are necessary, and themselves) back. The authors believe in well be effective, but I believe other tech- may well be effective. making high-quality group participation niques can also be effective and ulti- I invite you to read the book and make a conscious goal. It is not necessary to be mately less invasive. up your own mind. It will make you fully engaged all the time. But it is There are many examples in our society think and may challenge your assump- required to be honest with yourself and of cooperation at a much higher level tions; even the ideas that you might not your coworkers about whether you are than most software teams achieve. accept are interesting and strangely “in” or “out” at any time. Sports teams are an obvious example. A compelling. And they grow on you.

58 Vol. 27, No. 3 ;login: