[ih] Modèle OSI : Fwd: FW: internet-history "Digest, Vol 11, Issue 1"

[John Day]

At 9:09 +0100 2005/12/08, D. Battu wrote:
>Hi ! I am not really familiar with discussion list management, but 
>from what I heard here, I can tell you the few of what I remember 
>from the years 1970 and after.
>
>The references (*) from Pouzin, Zimmerman, and Marty are relevant 
>for me and these persons can  still be contacted directly (Marty is 
>still in function in FTR&D in Rennes). They certainly be pleased to 
>explain the own views !!!
>
>There was first a conflict between ITU and ISO/CEI about the 
>structure of information transfer in the 1970 (as far as I could 
>hear atthat time). Each entity had its own job and few cooperation 
>was possible between the two groups, so the scheme know as "ISO 
>model" came late after several long discussions (see also EURAS and 
>SIIT2005 web sites, and you could  ask MM. Sheriff and Kai Jacobs 
>about).

There was at least one draft of a model by CCITT SGVII before the two 
projects were made joint between ISO and CCITT.  The main difference 
was that CCITT tried to use the technology to define who owned what 
box.  For example, this is why an X.29 PAD was defined as part of the 
network rather than as a low function host.

>
>Secund, the main problem came, on the French side, from the 
>necessity to explain X.25 at a time where Fax and Minitel were 
>existing but not structured into layers !!!

Actually it did. V.21, LAPB, X.25 PLP, and the application.  The 
problem was that CCITT did not want to admit to the requirement for a 
Transport Layer.  A transport layer broke the "beads on a string" 
model where they could claim sole ownership of applications.  With a 
transport layer, they couldn't claim that only they could provide the 
application.

>Chronology suggests that the global view on the information layers 
>system has been a long work(probably ten years before some success). 
>In 1978, came the first draft of X.25 and the first contract for 
>Transpac (X.25 for France) and then Euronet (X.25 for Europe). In 
>these two contracts, the technical addendum was explaining how to 
>structure the information and the lawyers were upset when reading 
>such contractual implementation. (Because X.200 was not available at 
>that time, everything needed to be explain !). In 1979, early 1980 
>the X.75 standard has been published as a draft, the final document 
>been approved in the middle of 1980.

The first X.25 came out of the 1976 CCITT books.  It had major holes. 
It basically only described what the network had to do not the user.

>
>Then, the French administration decided to clarify the status of the 
>"information transfer" file and created a special group named 
>"Architel" (for architecture for telecommunications) in which were 
>working the persons I have mentionned in (*) - We spend a lot of 
>time and money for that, and we modified the Minitel in such a way 
>that it could comply with all the ISO model standards, which means 
>that 1983-1984 was the end of the Architel work and that probably an 
>agreement has been reach between ISO and ITU-T.

Yes, this was the theft of the Session Layer and the creation of TP0. 
This all came down at a WG meeting in Paris in the late summer of 83. 
It later turned out to be one of the big mistakes in the upper layer 
architecture, since the mechanisms in Session were not "re-entrant." 
Upon further reflection, I would have to check the archives but this 
may have been 82, not 83.  Because by the ISO 83 meeting, the work 
was much further along.  We already had sufficient drafts of Session 
protocols to fix them so that the upper 3 layers could be implemented 
as a single state machine.  (The birth of the clueless test! ;-))

>But the battle has been very hard between the delegations and 
>because of the acidity of the conflict, bad jokes have been 
>performed inside the standards and unfortunately heavy errors have 
>been accepted officialy inside the ISO/ITU papers which still 
>remains (heavy errors in X.25 mainly !). (Don't repeat it !)

The lesson of course is don't accommodate old technology.  This was 
true in the OSI work and also in the Internet work:  Just look at how 
many warts are still hanging around because of the TIPS!  ;-) What is 
really criminal is that some of those warts are now considered Best 
Practice!

>Thank you to you to not dispatch my answer around you - there is 
>still hostility in the air somewhere !
>(As we usually say in France " this one who say the truth should be killed !")

History should never be killed or we are doomed to repeat it.

Take care,
John