michaelo@tektronix.UUCP (Michael O'Hair) (05/18/85)
In response to Jon Shapiro's request and for general information. There are four groups involved in the OSI standardization: NBS, IEEE, ISO, and CCITT. Their addresses: National Bureau of Standards Building 225 Room B218 Gaithersburg, MD 20899 IEEE Standards Office 345 E. 47 Street New York, NY 10017 ISO documents may be obtained from: Frances E. Schrotter ANSI ISO TC97/SC6 Secretariat 1430 Broadway New York, NY 10018 CCITT documents may be obtained from: International Telecommunications Union Place des Nations CH 1211 Geneve 20 Switzerland Be forewarned that the volume of paper will be dangerous to your lower back! Michael J. O'Hair tektronix!michaelo Tektronix, Inc. P.O. Box 500 Beaverton, OR 97077 M/S 782
karn@petrus.UUCP (05/20/85)
While you're at it, you should give equal time to the loyal opposition. Michael Padlipsky has recently published a book through Prentice Hall entitled "The Elements of Networking Style" (ISBN 0-13-268111-0.) It is a compendium of critiques on the many questionable aspects of the ISO/OSI and CCITT approaches to networking. It presents a pretty strong case in favor of the ARPA protocols (including TCP/IP), by pointing out that they have already done for almost 10 years everything that ISO is just now getting around to re-inventing. Well worth reading if you'd like to avoid buying a picture of a pig in a poke (to quote one of the author's favorite expressions.) Phil Karn
mark@cbosgd.UUCP (Mark Horton) (05/20/85)
In article <354@petrus.UUCP> karn@petrus.UUCP writes: >Michael Padlipsky has recently published a book through Prentice Hall >entitled "The Elements of Networking Style" (ISBN 0-13-268111-0.) For those who dn't want to wait and have access to the ARPANET RFC's, you might want to read RFC's 87[12345], which were written by Padlipsky and probably address the same topic. My personal opinion, after reading these RFC's, was that while I agree with what Padlipsky is trying to say, he must have written those RFC's after a particularly heated argument with an ISO proponent. He argues so strongly and not especially coherently, and repeats himself so much, that it's hard to come away from the RFC's agreeing with him. Perhaps the book is better presented, having had more time to cool down. However, it wouldn't hurt to have a pair of asbestos gloves handy when you go to read it. His fundamental point, as I recall, was that since ISO has 7 layers, networking software will spend so much effort interfacing between layers that performance will go down the tubes. While I agree with this overall opinion of ISO, I must point out that, at least in theory, some of the layers can be null, or implemented by coding conventions and not with separate software. Mark
larry@anasazi.UUCP (Larry Rodis) (05/22/85)
> While you're at it, you should give equal time to the loyal opposition. > > Michael Padlipsky has recently published a book through Prentice Hall > entitled "The Elements of Networking Style" (ISBN 0-13-268111-0.) It is a > compendium of critiques on the many questionable aspects of the ISO/OSI and > CCITT approaches to networking. It presents a pretty strong case in favor > of the ARPA protocols (including TCP/IP), by pointing out that they have > already done for almost 10 years everything that ISO is just now getting > around to re-inventing. > > Well worth reading if you'd like to avoid buying a picture of a pig in a > poke (to quote one of the author's favorite expressions.) > > Phil Karn There is one problem with your suggestion to use ARPA protocols over ISO/OSI and that ISO/OSI is an international standard (or will be). That is the biggest reason to use it. I realize that it is not the best (I'm working on X.25 and the Transport Layer) but it is being molded into a standard. There is much to be said to writing systems around a standard that is being accepted worldwide. The most logical example is unix, it may not be the best OS in the world but it will run on more machines than anything else and therefore many companies are putting unix on there systems. Considering how much software maintenance costs portability is the name of the game, and the ISO OSI protocals will help to achieve network portability. -- Larry Rodis UUCP: {decvax|ihnp4|hao}!noao!terak!anasazi!larry ucbvax!arizona!asuvax!anasazi!larry PHONE: +1 (602)275-0302
jst@wucs.UUCP (Jon Turner) (05/30/85)
In article <198@anasazi.UUCP> larry@anasazi.UUCP (Larry Rodis) writes: >> >> Michael Padlipsky has recently published a book through Prentice Hall >> entitled "The Elements of Networking Style" (ISBN 0-13-268111-0.) It is a >> compendium of critiques on the many questionable aspects of the ISO/OSI and >> CCITT approaches to networking. It presents a pretty strong case in favor >> of the ARPA protocols (including TCP/IP), by pointing out that they have >> already done for almost 10 years everything that ISO is just now getting >> around to re-inventing. >> > >There is one problem with your suggestion to use ARPA protocols over ISO/OSI >and that ISO/OSI is an international standard (or will be). That is the >biggest reason to use it. I realize that it is not the best (I'm working on >X.25 and the Transport Layer) but it is being molded into a standard. >There is much to be said to writing systems around a standard that is >being accepted worldwide. The most logical example is unix, it may not be >the best OS in the world but it will run on more machines than anything else >and therefore many companies are putting unix on there systems. >Considering how much software maintenance costs portability is the name >of the game, and the ISO OSI protocals will help to achieve network >portability. > I'm afraid you picked a terrible analogy. Unix was a very successful operating system with a lot of miles on it before the standardization effort began. That is, it was a de facto standard before anyone thought of making it an official standard. (So far as I know, it's still not an official standard but I don't think the lack of standardization will in any way impede its success.) The ARPA protocols are also a de facto standard. They've been around for a lot of years and have a lot of mileage on them in a wide variety of different machines and operating systems. Many places uses them, including anyplace that runs 4.2 BSD. The ISO/OSI protocols are not a de facto standard and may never be. One of Padlipsky's big questions is (to paraphrase) "Why commit your future to a committee-designed protocol suite, which has never been fully implemented, with which there is limited operational experience and which has a lot of known warts when there exists a very workable de facto standard already available?" I think he's got a point. I second Phil Karn's recommedation of Padlipsky's book. It's one of the funniest books I've ever read and is worth reading just for entertainment. It also has a serious message which should be heeded by anyone getting into the "intercomputer networking game." -- Jon Turner Washington University in St. Louis 314-889-6193 UUCP: jst@wucs.UUCP or ..!{ihnp4,seismo}!wucs!jst ARPANET: wucs!jst@seismo.ARPA CSNET: wucs!jst@seismo.ARPA%csnet-relay