Summary

  • In 1986, Spanish representatives joined a RARE meeting in Copenhagen for a network that a participant later called a “network on paper”. A 1987 Valencia workshop tested the idea with two X.25 lines and four borrowed PCs; it was not a national service.
  • The National R&D plan and Programme IRIS formally began on 1 January 1988. CICYT funded and supervised the programme, while Fundesco managed it. Its early architecture favored open ISO/OSI protocols over public X.25 service.
  • University demand for TCP/IP and the gap between OSI plans and working implementations helped redirect the route. SIDERAL/IP was tested in 1990, with four experimental centres connected by year-end; the operational SIDERAL service began in March 1991.

A European meeting before a Spanish network

The first RedIRIS story is often told from the name the network eventually acquired. That can make the beginning sound more complete than it was. Ramón López de Arenosa, a participant in the programme, later recalled that planning for a Spanish academic network was under way from 1985. In spring 1986, Spanish delegates went to a RARE meeting in Copenhagen as representatives of what he remembered as a “network on paper”. Spain offered to host a subsequent meeting, which led to a workshop in Valencia in autumn 1987.

The distinction is more than a rhetorical flourish. Spain had a place in a European conversation about research networking before it had a national network able to serve universities. The delegation represented an intention and a project in formation. It did not demonstrate that Spanish campuses could already exchange traffic with one another or with European peers.

The Valencia workshop made the gap concrete. López de Arenosa recalled two X.25 lines and four borrowed personal computers. Those resources were enough to demonstrate a direction and bring people together; the account does not describe a countrywide service, a measured user base or continuous connectivity. The technical experiment and the institutional preparation were related, but they were not interchangeable evidence.

A programme gave the plan an operating home

The formal step came on 1 January 1988, when Spain’s National R&D plan and Programme IRIS began. CICYT supplied funding and supervision; Fundesco was selected to manage the programme. Fundesco later coined the name RedIRIS. This order matters: the programme began in 1988, while the familiar network name and services developed across subsequent years.

IRIS’s early design favored open protocols associated with ISO, OSI and CCITT, with public X.25 service over Iberpac as an important transport layer. The choice made sense in the European policy setting of the period. RARE and COSINE were building a framework for cooperation among national research networks, and open standards promised an interoperable alternative to incompatible proprietary systems.

The programme was also more than protocol selection. Its managers supported institutions that lacked connectivity with X.25 circuits, interface cards and software, and helped existing networks migrate when they could reach an agreement. A RedIRIS institutional history reports that more than 200 institutions were connected by the end of 1990. That is a count of institutions in the programme’s account, not a count of individual users, equivalent service levels or full Internet connections.

Running services changed the architectural balance

ARTIX supplied early OSI services during the 1990–93 period. But implementation did not advance as quickly as the plan anticipated, and universities were asking for TCP/IP, which was already proving useful in working networks. The retrospective history describes this as a practical mismatch: the programme had a coherent open-systems design, while users needed a protocol suite they could deploy with the equipment and connections available to them.

RedIRIS’s account dates Spain’s first full Internet connection to an experimental link in mid-1990. By the end of that year, four experimental centres were using IP over X.25 through the 64 Kbps IXI network. SIDERAL/IP was the service name associated with that turn. RedIRIS later records operational SIDERAL service from March 1991. These are separate milestones: an experimental international connection, a small set of pilot centres and an operational national service should not be collapsed into one “launch”.

This is where the history becomes useful beyond Spain. The initial OSI direction was not irrational, and the sources do not show that participants simply abandoned it overnight. But the architecture had to meet the constraints of working institutions. Once universities asked for TCP/IP and could use it, the operational path shifted. In Heng Lu’s later Note 65, implementation experience is a discipline on architectural claims; Note 64 adds the distinction between a common specification and the choices participants actually adopt. These notes are analytical lenses written decades later, not evidence of what Spanish officials or RARE intended.

What “started” means

There is no single date that captures every beginning. Planning preceded the 1988 programme. European representation preceded national operation. The Valencia workshop tested a small configuration. The 1990 experimental connection did not equal full operational service. March 1991 marks the start of operational SIDERAL in RedIRIS’s retrospective chronology.

That sequence also sets a limit on the story. The available detailed accounts are institutional retrospectives published in 1998, alongside later official timelines. They establish a useful chronology and describe the programme’s own explanation for its technical turn, but they are not contemporary network logs. They do not supply packet traces, traffic volumes, comparative costs or a complete record of how each institution made its choice. “Demand for TCP/IP” and “OSI implementation lag” are supported explanations in RedIRIS’s account; their relative causal weight cannot be calculated from the material reviewed here.

Still, the sequence corrects a common shortcut. A country can be represented in network coordination before it operates the network being discussed. A programme can formally exist before it provides nationwide service. And a standards strategy can give way when institutions find that another protocol path is usable. The name RedIRIS eventually covered a network built through these separate steps, not a single founding moment.

Sources