The protocol war

In retrospect, the year 1988 in many respects was decisive for the future of the networking scene in Europe, and thus between Europa and the rest of the world. It was in that year that the "protocol war" culminated and led to the separation of the "official bodies" (European Commission (CEC), national governments, PTT's, RARE) and the "practical workers" (Eunet, EARN, CERN/HEPNET).
 
In 1985 a project called "COSINE" (Cooperation for OSI Networking in Europe) was adopted as part of the Eureka programme at the 2nd Eureka Ministerial Conference in Hannover. The objective of the project was to provide the European academic and industrial research community with a computer based open communications infrastructure. In COSINE, all Western European countries amd the European Commission participated. Thus, COSINE was the "most European" project in Eureka. COSINE was set up to rapidly establishan environment to make data communication services available, based based upon "International ISO Standards and CCITT Recommendations for Open Systems Interconnection (OSI) together with functional standards as prepared by CEN/CENELEC, CEPT, ETSI and EWOS". The 3-year implementation phase was to start in January 1989.
 
On April 25, 1988 a committee, the "Joint Working Party A" (WP-A), comprising representatives of "officials" and "networkers" was formed under the aegis of COSINE, to make a proposal to achieve before 1989 the common use of an X.25 (1984) infrastructure ("IXI") for all international data traffic of the European R&D community. Despite several meetings of this WP-A, consensus was never reached. The most fierce opposition, in fact boiling down to a veto, to all plans of the "networkers" in the group where they even remotely dared to mention non-OSI stuff like IP, came from the EC.
 
Below is a comment that I, as a participant in WP-A, wrote up for EUnet. Shortly thereafter EUnet decided to go its own way and embrace IP, since it considered that the only way to ensure fast, large-scale and future-safe local and (inter)national networking, in particular with the Internet/NSFnet in the USA, without the need for clumsy gateways and protocol converters (with which we already had some experience). And within just a few months, in november 1988, CWI/EUnet had its own direct connection to the Internet.
 
This CWI/EUnet internet connection put the Dutch national research network, SURFnet, in a very difficult position, since EUnet could start offering IP service, including to academic institutions, which SURFnet was supposed to do. But SURFnet, being 50% owned by the Dutch PTT and sponsored by the government, had to adopt OSI, although it too realized that IP was the way to go. Faced with this dilemma, it didn't take long before SURFnet took a pragmatic approach by formally allowing only OSI on its network, but informally IP too. Problem solved.
 
The COSINE network was indeed established, but this expensive mistake was short-lived, and IP has taken over the networking world, from high-level R&D down to the consumer level.

Some abbreviations used here:
• EUnet: European Unix Network
• EARN: European Academic Research Network (IBM-sponsored)
• HEPNET: High-Energy Physics Network
• CPG: COSINE Policy Group
• MDNS: Managed Data Network Service (by the PTT's)
• IXI: International X.25 Infrastructure (PTT's) 


Date: 30-Jun-1988
Subject: WPA/CPG

Here are some (personal!) comments on the interim report produced by
Working Party A and the decisions made by the Cosine Policy Group.
Furthermore an overview of current developments and activities around
a shared pan-European X.25 infrastructure.


Comments:

  The whole WPA business was a tough thing, with almost continuous
  clashes between the two "main parties" involved, i.e. EUnet, EARN
  and HEPNET on one side, COSINE, RARE and CEC/IES on the other. It
  was only after wasting most of three (!) full meetings that things
  were more or less settled.
  Almost right from the start it was clear what RARE/COSINE/CEC had
  in mind: get an infrastructure off the ground on the shortest possible
  term, managed by the PTT's MDNS (only a CEPT project at the moment,
  not a real service or company!). Most strongly opposed to this was
  EARN, with its own complete migration strategy and new infrastructure
  to be set up *very* soon. But neither EUnet nor HEPNET were in any
  way happy with the idea of PTT's managing a European infrastructure,
  the same PTT's that sofar have shown not to be able to provide an
  international X.25 infrastructure that even remotely comes close to
  the needs of the existing networks; just look at the effective speed
  of uucp (not to/from mcvax of course....) over international X.25
  links and you'll know what I mean.
  In the last WPA meeting, prior to the CPG meeting, a complete draft
  of the interim report was drawn up and discussed. In the report
  two possible implementations of an X.25 infrastructure were by then
  proposed; EUnet and HEPNET (EARN was not represented at that meeting)
  objected to these and added a third possibility, that in the final
  report became option 1: an upgrade and migration of the current
  networks (EUnet, EARN, HEPNET) into a shared X.25 infrastructure.
  Note that the wording of option 1 in the report also allows reading
  "the PSDN's" for the "existing networks"; politics of course.
  Since - as said - it was clear that the RARE/COSINE/CEC "group" would
  strongly favour the MDNS solution and since it could be expected that
  this solution was to be presented as the best solution to the CPG, I
  wrote up a complete proposal, representing the viewpoint of EUnet.
  The draft was discussed with [...] (EUnet) and [...] (HEPNET) and
  faxed to [...] (EARN), all on the Friday before the CPG
  meeting; after some changes all gave their consent, so the proposal
  could be presented as backed by EARN and HEPNET. It was added as an
  official appendix (Appendix D) to the report.

  If you look at the decisions made by the CPG, you'll see that they
  indeed chose for the MDNS solution (option 2 in the report), as could
  be expected. After I heard of that, my immediate reaction was that
  EUnet, EARN and HEPNET should stop participating in WPA and instead
  join efforts to set up their own shared infrastructure. But after I
  got the complete text of the CPG decisions and heard (from [...],
  the person with the many hats [RARE, COSINE etc.]) about the background
  of the decisions, my view changed. See below.
  Important reasons for the decision were:
  - until the last minute Germany has always heavily opposed to the
    Statement of Direction and Cooperation; the CPG gave in and therefore
    that Statement was never officially ratified. At the last CPG meeting
    Germany demanded that all participants in the new infrastructure would
    be allowed to **only** run ISO/OSI protocols over that infrastructure.
    This time the other CPG members did not give in and Germany was left
    alone; the main argument was that continuity of existing services
    should be guaranteed.
  - it was clear that most countries represented in the CPG would not
    accept it if the PTT's were not actively involved (other than being
    the supplier of the lines) and would only go along with a CEPT/MDNS
    solution.
  An absolute requirement from EUnet, EARN and HEPNET has always been
  that there should be no restrictions on running specific (existing)
  protocols over the new infrastructure; as said, this requirement was
  met by the CPG decisions. An additional requirement from EUnet was that
  the *whole* current EUnet user group should be given continued service,
  independent of the nature of a specific EUnet site. Now, if you read the
  text of the CPG decisions, you'll see that this EUnet requirement was
  indeed met in point 6: EUnet is indeed a network "supporting research
  in Europe".


	Piet