Summary

  • BELNET’s first network followed a reported decision by academics and science-policy officials to create one national network for communities then organized around EARN and EUnet.
  • The plan and the network’s first 64 Kbps connections are not proof that every host, route or service migrated in 1993; RIPE still listed Belnet and EUnet BE separately as providers that November.

A network decision before a network

In November 1993, a draft list circulated by the RIPE Network Coordination Centre named two Belgian Internet service providers: Belnet and EUnet BE. The entry records registry roles, not the route between two campuses. It cannot show whether the providers shared equipment, exchanged traffic, or used the same backbone. But it offers a contemporaneous snapshot against which a later institutional account can be read.

Belnet’s 2008 annual report looks back on the work that preceded the network. Philippe van Bastelaer, then an emeritus professor at Namur, recalled a fragmented landscape and a divide between academics aligned with EARN and EUnet. He said an informal expert meeting agreed to create a single new network to which both existing networks would migrate. A working group of academics and officials from the Federal Science Policy Office, led by Pierre Bruyère, then set up an official expert group to determine the future network’s architecture and transmission capacity.

That account describes a design decision. It does not give the meeting date, a migration schedule, or a list of systems that had switched over. The RIPE list supplies a different kind of evidence: in November 1993, both Belnet and EUnet BE still appeared under their own provider entries. Neither source cancels the other. A shared research backbone could be under construction or in service while EUnet continued to operate separately for some users or functions. The available records do not settle exactly how those layers overlapped.

Two communities, with a bridge between them

The communities had different institutional homes and computing cultures. In a 1999 KU Leuven interview, professor Pierre Verbaeten described EUnet Belgium as managed from Leuven’s computer-science department and associated with Unix systems. Herman Van Uytven described EARN/Bitnet as a network of large IBM mainframes, with Leuven’s computing centre at the heart of its Belgian activity. Both recalled email, file exchange and other academic uses.

The same interview cautions against a simple story of total isolation: Verbaeten said EUnet Belgium provided mail contact between EUnet and EARN/Bitnet. Interconnection at the level of a service did not necessarily make the networks one operational system. Users could exchange mail across a boundary while institutions still maintained different hosts, lines, administrative arrangements and ideas about the next network.

A separate motive appears in Belnet’s official history. In 1989, the Science Policy department established its Information Technologies Impulse programme to encourage Belgian researchers to use supercomputers. One strand concerned a network that would let researchers access those machines remotely. The record therefore places network-building inside a research-computing programme, not just a contest to claim the country’s first Internet connection.

What 64 Kbps can—and cannot—say

Belnet says its first network dates from 1993. The annual report states that its first customers were connected to one another at 64 Kbps. A 2023 anniversary account reports three employees, 27 connected institutions and the same capacity at the organization’s founding. These are useful measures of reported early scope. They do not establish a nationwide consumer service, equal access for every institution, traffic volumes, or a performance audit.

KU Leuven’s retrospective adds a transition detail: several Belgian universities began connecting to the Internet through Leuven from 1990, and, in 1993, the national and international connections were transferred from Leuven to Brussels as Belnet was founded. It is a participant’s recollection of a specific shift in the location of connections. It does not prove that every EARN and EUnet service moved with them.

A 1996 Belgian legal analysis later described a Brussels interconnection router managed by Belnet and explained how peering could keep domestic traffic from taking international detours. By then, national research networking and provider interconnection were related but distinct jobs. That later description should not be projected backward as a precise map of the 1993 backbone; it does show why a network operator could be a common meeting point without absorbing every provider.

The boundary of the evidence

BELNET’s founding story is not merely that a 64 Kbps circuit appeared. It is that a mixed group tried to replace fragmentation with a shared design and a migration path. Yet the record does not show that a decision instantly dissolved existing networks. The November 1993 RIPE entries are especially useful for keeping those statements apart: the common-network plan is retrospective testimony; the two provider entries are a contemporaneous registry snapshot, incomplete by RIPE’s own warning.

The strongest conclusion is narrow. Belgian institutions had begun building a common research-network layer, with an explicit aim to bring earlier systems into a shared architecture. The operational boundary remained plural at least in the provider record. Whether a particular university had migrated, retained an EUnet service, or used both cannot be read from those entries alone.

The distinction matters in Internet history. A committee can agree on architecture; operators still have to move lines, hosts, routes and users. In Belgium, the agreement is part of the story, not evidence that the transition was already finished.

Sources