Summary

  • Iran’s first cross-border academic link carried email through BITNET and EARN; the later TCP/IP connection was a separate technical step.
  • The record disagrees on the first date and Austrian endpoint, while a 1996 snapshot shows that an Internet node still depended on slow international and local links.

“Connected” names more than one milestone

Iran’s Internet history is often assigned a single arrival year. The surviving accounts describe two different events: academic email through BITNET and the European Academic and Research Network (EARN), then a transition to TCP/IP. A mail exchange, an IP-capable node and a service available to ordinary users are different forms of connectivity.

The effort grew out of research needs. Siavash Shahshahani, then deputy director of the Institute for Studies in Theoretical Physics and Mathematics (IPM), recalled in an interview that the institute wanted researchers to stay in contact with universities abroad. He said Abdus Salam’s International Centre for Theoretical Physics helped introduce IPM to EARN. Shahshahani remembered an initial telephone connection to the University of Linz late in 1992, followed by a leased line to the University of Vienna.

The 1999 Open Research Network survey instead dates a 9,600-baud line to Vienna to January 1993 and says it first carried BITNET/EARN traffic. The sources do not settle the first date or Austrian endpoint.

The same caution applies to “Internet.” Shahshahani recalled IP operation late in 1993, with Iranian addresses initially appearing under Vienna’s UNIVIE.AT namespace. The 1999 survey places the TCP/IP gateway change in mid-1994: IPM adapted a MicroVAX gateway for IREARN members, still using the 9.6-kbps line to Vienna. It describes a Class C allocation, but its chapters differ between 500 and 512 addresses. The number is uncertain; the architectural change is clearer. IPM’s academic gateway could carry TCP/IP, and universities reached the outside network through links to IPM.

An April 1994 network-administration record lists an .ir domain, nameservers and several IPM hosts, but an accompanying message says connections appeared unavailable that morning. This is a dated observation, not proof that all Iranian connections were down. It shows why a registered domain or listed host cannot alone establish reachability.

By 1996, the network had widened but remained uneven. A contemporary survey describes two 9,600-baud lines between IPM and Vienna, mostly 9,600-baud domestic academic links, and only two 64-kbps links in the internal network it surveyed. It reports congestion and routing problems on the Vienna path. Its estimate of tens of thousands of users is not a household census; many of IPM’s 1,100 server usernames were shared institutional accounts. The survey also says a satellite link with potential 2-Mbps capacity was technically ready but still awaited DCI–Intelsat paperwork. It does not establish why or when the link entered service.

There was no single “arrival” switch. Email crossed borders first; TCP/IP followed; local routes, capacity and commercial services developed on their own timelines. Treating them as one moment makes the connection look broader than the evidence allows.

Sources