Summary

  • A November 1993 operator report describes Ariadne as a working X.25 research network with dial-up, public-data-network and institutional access; its roughly 1,000 users were an estimate, not a national census.
  • The same report calls the eight-site router build an experiment under installation. That distinction matters: international Internet reachability, an operating X.25 network and a completed national IP backbone were different states.

Ariadne’s history is easy to flatten into a protocol victory: Greece chose TCP/IP, the old network disappeared, and a national Internet emerged. The surviving records describe something more gradual. In November 1993, the people running ARIADnet reported an active academic and research service built primarily on private X.25 links. They also described an IP experiment that was still being installed. The difference between those two sentences is the history.

The programme grew from Greece’s research and technology administration and, by mid-1988, was administered by the National Centre for Scientific Research “Demokritos” in Athens. Its purpose was institutional reach. Universities, research centres and smaller research-and-development groups could connect their local machines and let users reach hosts away from their campus. The network was not just a cable map: it combined telecommunications circuits, computer centres, operating staff, access arrangements and help for people learning what the network could do.

The 1993 report gives a useful, bounded snapshot. It calls Ariadne a private X.25 network and lists three access routes. Seven organizations subscribed through HELLASPAC, the public data network; seventy used Ariadne’s dial-up service; and the report counted about one hundred organizations served in all, with an estimated thousand end users. Those categories do not reconcile neatly into a verified register, and the user figure is explicitly an estimate. They still show that the service reached beyond a handful of central university computers.

Small departments and research groups could use a dial-up entrance rather than buying a permanent circuit of their own.

International connection also came through more than one path. Ariadne linked to HELLASPAC through two 9.6-kilobit-per-second X.25 links, to IXI through one 9.6-kilobit link and another 64-kilobit link that the November report said would soon become operational, and to Greece’s EARN node in Heraklion, Crete. The report says a CERN IP link was about to be established. It does not say that the planned link was already carrying traffic. Nor does the presence of an Internet path prove that every Ariadne service or attached site had become a native IP network.

Ariadne’s operating surface was deliberately mixed. Its services included X.400 mail, Triple X, Internet IP and DECnet over X.25. The Demokritos operations centre had a central router, while routers with protocol-translation capabilities were planned for major research institutions in the north, south, east and west, as well as Athens. Eight sites were being connected in an experimental router network intended to offer higher-speed IP service where Internet IP had previously been offered over X.25. This was a bridge architecture: preserve access to several existing systems while testing a different packet service.

The report’s future tense matters. It records an installation programme, not a finished eight-site conversion.

People and information services formed part of that infrastructure. The Demokritos network team operated the NOC and handled a growing workload across varied vendors and operating systems. Ariadne offered a help desk by phone, email and network dialogue, and its Pythia server let users browse information about the network and related subjects. A chemists’ support group helped users discover databases, software, FTP sites and discussion lists, then documented what it learned. The report even describes work on Greek-character handling for terminals.

Access depended on making the machinery usable in the language and work practices of its researchers.

A later peer-reviewed history places the change from OSI to TCP/IP between 1992 and 1995 and says GRNET emerged as a research project in 1995; a Ministry-backed organization with the same name followed in 1998. These milestones should not be compressed into one “launch.” They describe a design transition, a project and a later institution. A 1998 technical study then depicts GR-NET across six Greek cities and records a 10-megabit-per-second ATM connection from Athens to Milan. It reports 121,000 users hosted at institutions at the beginning of 1999, compared with 289,000 total Internet end users in Greece.

That was a substantial research network, but those numbers do not mean that GR-NET served every Greek Internet user or household.

The useful historical question is therefore not simply which protocol won. Ariadne made several kinds of access and several network systems work together while a newer IP design was being built. Its later transition moved the common direction of the research network; it did not retroactively make the earlier X.25 service unreal. The record shows a series of operational choices, each with its own evidence: a circuit installed, a gateway reachable, a router experiment underway, a service used, and an organization formed.

Sources: ARIADnet operator report, November 1993; Greek Informatics: A Critical History (2018); Greek academic and research network case study; University of Athens Computing Center chronology; National Documentation Centre, 1990s milestones; GRNET institutional history and profile.