Summary

  • The Vienna–CERN fixed circuit entered service in February 1990 using IBM SNA; a participant’s later account dates its TCP/IP operation to 10 August.
  • ACOnet’s national university network began as a separate X.25 ring and migrated to an IP backbone in 1992.

The phrase “Austria connected to the Internet in 1990” compresses several different histories into one date. It can mean that Austrian researchers used international networks, that a leased circuit reached a foreign research center, that the circuit carried TCP/IP, or that a national backbone itself used IP. The sources do not make those events simultaneous.

A first distinction predates the famous date. The University of Vienna’s institutional chronology says Austrian universities gained access to international data networks such as EARN and Usenet in 1985. Peter Rastl, who led the university computing center, later recalled that people in Austria had some Internet access by dial-up before a permanent line. These statements describe different services and should not be collapsed into a claim that Austria had a national Internet backbone in 1985.

The fixed Vienna–Geneva circuit also had more than one start. In a 2010 interview, Rastl said the line was requested from the Post and Telegraph Administration on 12 June 1988 and approved on 31 October. It entered service on 10 February 1990, but initially used IBM’s SNA network protocol to reach the CERN computer center. The line’s physical or leased operation therefore preceded its use as a TCP/IP connection.

Rastl dated the protocol change to 10 August 1990. The circuit, named “Geneve-Wien NP1,” then ran TCP/IP between the University of Vienna’s IBM mainframe and CERN. The Vienna segment was 64 kilobits per second. CERN, in turn, had a 1.5-megabit-per-second T1 connection to the US NSFNET. The faster number belonged to CERN’s transatlantic link, not the Austrian circuit. The path mattered because it joined a local university host to an international research network; it did not make Vienna a direct endpoint on the US line.

At the same time, ACOnet was building a different layer. The Austrian Academic Computer Network’s first nationwide infrastructure was a private X.25 ring linking university sites in Vienna, Graz, Leoben, Klagenfurt, Innsbruck, Salzburg and Linz. Its connections started at 9.6 kilobits per second and later rose to 64 kilobits per second. ACOnet could connect to Internet and other research networks while its own national carrier network still used X.25. “Connected to the Internet” and “the domestic backbone runs IP” were not interchangeable descriptions.

That distinction lasted into 1992. ACOnet’s official history dates a migration from X.25 to IP to that year. The new core formed a triangle among Vienna, Linz and Graz, with other sites attached to those nodes; the lines were then 64 or 128 kilobits per second. The University of Vienna’s chronology separately says its computing center formally took responsibility for ACOnet in 1992. A 1992 status report by Walter Kunft still described the Austrian academic network as a private X.25 system and discussed a multiprotocol upgrade supporting IP, ISO CLNS and X.25.

Together, the records show a transition underway, not one nationwide protocol flip at a single instant.

A 1996 retrospective by Rastl adds an economic and regional layer. It says the international connections were raised to 256 kilobits per second in 1992 and describes later cooperation with Central and Eastern European research networks. RIPE’s 1992 Prague report records an ACOnet proposal to provide dual connectivity through Vienna and Linz to Bulgaria, Czechoslovakia, Hungary and Poland. That was a proposal discussed in a regional coordination setting, not evidence that every proposed circuit was already operating.

The precise milestone therefore depends on the question. The TCP/IP date marks the operation of a fixed international circuit. The X.25 ring marks an early shared national academic infrastructure. The 1992 migration marks ACOnet’s move to an IP core and a change in its operating arrangement. A single “arrival” date is useful shorthand only when its layer is named.

Sources

The article’s reading lens draws on Heng Lu, Note 65, “Running-Code Primacy” and Note 64, “Minimum Initial Specification, Localized Future Decision, and Voluntary Adoption”. These notes are interpretive references, not sources for the Austrian chronology.