historical pieces on scaling, tcp/ip, ...

I suppose the first time I even heard the term TCP/IP was around 1980, then represented as an up and coming replacement for ARPANet ... (alan j. maitland)

I've always thought of TCP/IP as an odd moniker. If it really meant "TCP over IP" it would not include IP alone, or UDP or RTP. And if it meant "TCP or IP" (like he/she) then it wouldn't be parallel - you'd expect TCP/UDP/RTP/ICMP to be the name ... (david p. reed)

the discussion about running out of internet addresses, is as old as 1988. vint cerf informed jon postel 1988 ...

also this sounds familiar

I want to see a protocol address space large enough to handle a node in each household appliance, each piece of electronic equipment, and several extras per household, office and vehicle. Traffic lights on the Internet. Stray toasters. And enough addresses left over to scatter hosts across the inner solar system. (john romkey, 30 nov 88, ToasterNet)

amazingly ietf meets this year en viena. it's a shame, missed ripe en barna, (webcast-archive) so maybe worth to think about a u-turn next month.

once upon a time there was a single file: HOSTS.TXT, maintained by SRI's (Stanford Research Institute in Menlo Park) Network Information Center.

then came tcp/ip, an explosion and

'... a host of problems with HOSTS.TXT': Traffic and load, Name collisions, Consistency. The essential problem was that the HOSTS.TXT mechanism didn't scale well. Ironically, the success of the ARPAnet as an experiment led to the failure and obsolescence of HOSTS.TXT.

Paul Mockapetris, then of USC's Information Sciences Institute, was responsible for designing the architecture of the new system. In 1984, he released RFCs 882 and 883, which describe the Domain Name System. These RFCs were superseded by RFCs 1034 and 1035, the current specifications of the Domain Name System.[2] RFCs 1034 and 1035 have now been augmented by many other RFCs, which describe potential DNS security problems, implementation problems, administrative gotchas, mechanisms for dynamically upd.' (dns&bind)

paul mockapetris wrote the first implementation of the dns, called jeeves. later came bind by kevin j. dunlap.

dns glossary from men and mice.
isc, internet software consortium

i knew a little bit about the quarrel: osi vs sna from ibm vs tcp/ip. but i haven't interpreted that fact so far to understand the european commission as a "promotor" of open systems. but it seems the european union had decided in the 80s not just to use osi, but also to support unix, at that time still in the hand of at&t.

osi (open systems interconnection) played one part in the game to push european computing business, but it didn't work out. well, also ibm lost that battle and tcp/ip finally won, without really playing the game, but just by being already there and - most importantly: functioning. jacques e.c. hymans case study on "open systems in the computer industry" deals with both stories: he describes how it came to osi and how to the desicion to use unix. (free access on ciao available for 30 days)

"Over dinner at the 1984 Cebit computer show in Hannover, Germany, the heads of the five biggest indigenous European vendors agreed to form a group dedicated to the purpose of providing a “common applications environment” based on Unix. At first, the group was called BISON (for Bull, ICL, Siemens, Olivetti, Nixdorf); when Philips joined the group, BISON changed its name to X/Open." (hymans, do too many chefs really spoil the broth? the european commission, bureaucratic politics and european integration, 1999)

another source for that at the bull's alumne site: "1984 Foundation of the X/Open Group by Bull, Nixdorf, Olivetti and Siemens to prepare common standards, based on UNIX. The Bull's intent was to have CEE (European Union) make those standards mandatory and consequently regulate the IBM de facto monopoly. The goal was also to limit the freedom of UNIX evolution by AT&T labs. The X/Open group was soon open to U.S. and other Companies and the strategic role decreased. Bull's active role in X/Open lead to its solicitation by IBM to found Open Systems Foundation."

well the french left out the english - as always, it seems. i just recall a conversation with raymond steele who was/¿is? sitting in one or another eu-board, that discussed the strategy for "how to go on" with umts. anyhow ICL of great britain - now fujitsu - was with the x/open group.

cooperations of european vendors was not by any means natural, heymans write. the first experience in this regard was made in the 70s by philips, siemens and the french "compagnie internationale pour l'informatique (CII). they formed unidata. but the dutch/german/french collaboration failed already in the same decade.

"Kompromisse waren an der Tagesordnung, was zu allseitiger Verärgerung führte. Es fehlte an der gebotenen Identifikation mit den vereinbarten Gemeinschaftsprojekten und mancher räumte seinen Eigenentwicklungen weiterhin Vorrang ein. Die Zusammenarbeit endete dann auch bevor sie Früchte tragen konnte, als auf Beschluss der französischen Regierung das gesamte inländische DV-Potenzial Frankreichs in der neu zu gründenden Gesellschaft CII-Honeywell-Bull konzentriert wurde. Die CII mußte also aus dem UNIDATA-Verbund ausscheiden. Damit war der Versuch, die Entwicklungskosten auf mehrere Schultern zu verteilen, gescheitert." (computermuseum münchen)

"european vendors primarily looked for overseas rather than intra-European alliances"(Mytelka and Delapierre 1987)

have to find those articles from h. landis gabel. book: Technical Standards and Competition in the Microcomputer Industry, in H. Landis Gabel (ed.), Product Standardization and Competitive Strategy, Elsevier Science Publishers B.V., Amsterdam, 1987.

a guide to X/Open