CSNET -- Computer Science Network


Sat Jan 16 10:29:05 1982
New CSNET Phonenet sites
From: Dave Farber

Two new sites are now accessible via Phonenet as part of CSNET. They are the University of California at Irvine (UCI) and the first of the industrial sites — the Hewlett Packard Laboratories (HP-Labs)… The current list of CSNET Phonenet sites is:

Udel ( Delaware )

Additional Phonenet sites will be connected in the near future. These will include New Mexico Tech, Univ of Penn, Georgia Tech, Duke/UNC, Fairchild, and the Univ of Md. It is expected that all sites with applications filed before 1 Nov will be online by mid March…

The Computer Science Network (CSNET) helped introduce what was fast becoming the Internet to universities around the world, and laid the groundwork for development of the NSFNET.

The history of the CSNET is a story of enterprising academic entrepreneurship, and yet another example of TCP/IP‘s inexorable drive to spread. In 1979, most U.S. universities weren’t doing research with the Department of Defence and so weren’t connected to the ARPANET, but were increasingly aware of the network’s advantages and wanted to level the research playing field. With seed funding and support from Kent Curtis at the National Science Foundation, Larry Landweber at the University of Wisconsin-Madison put together a proposal to build a network to connect non-ARPANET computer science departments. The proposal made its way to Dave Farber for review, who gave it to one of his graduate students, Dave Crocker, who thought it was an interesting idea but, like others, worried about the university’s lack of networking experience.

With the support of Vinton Cerf, DARPA, and others, the NSF awarded $5 million to the CSNET project in January, 1980. A team was assembled to manage the project — Landweber, Farber, Peter Denning, Tony Hearn, and the NSF project manager Bill Kern. One of the immediate consequences of the award was the connection of three universities to the ARPANET that became the core of the CSNET — Wisconsin-Madison, Purdue, and Delaware. In a foreshadowing of the NSFNET development, one of the conditions of the award was a key clause requiring the network to become self sufficient within five years.

To progress the plan to develop the CSNET, a meeting was held in the Philadelphia airport between representatives from DARPA, the NSF, and several universities. The meeting started at 9:00 in the morning and went on for several hours, during which the group gradually realized how limited the commercial possibilities were for establishing a network with the advanced functionality provided by the ARPANET. At around 3:00 in the afternoon, Crocker suggested that a simple telephone-based email relay was likely feasible, similar to the MMDF service he had established for the U.S. Army. The group liked the idea, and asked Crocker how long it would take to build. He gave them a reasonable theoretical estimate — one day.

This email relay network became known as Phonenet and got the CSNET off the ground, but took somewhat longer that a day to actually implement. The MMDF software was extremely difficult to install, and needed to be scaled to support a wide range of new sites. In something of an understatement, Crocker recalls their experience with one of the first connections: “One of the early sites was attached to Telenet so we ran the dial-up over it. That is, we did telephone emulation over an X.29/X.25 network. It was an interesting lesson in the inefficiencies of mismatched packet sizes”.

By the end of 1981, the University of Delaware, Princeton University, and Purdue University were connected to Phonenet, using Delaware and RAND as relay sites to the ARPANET. By 1982, there were 24 sites connected, and by 1984 there were 84 sites, providing email connectivity for university computer science and engineering departments around the country. Eventually a host was also established at Wisconsin-Madison to provide dial-in and network access to the ARPANET as well.

To provide full internetworking functionality, Vinton Cerf proposed establishment of a network gateway between CSNET and the ARPANET, and naturally TCP/IP was chosen as the common network protocol. To ensure that it remained the standard, TCP/IP was made available for free to CSNET sites, and so became the common standard across the network. CSNET subsequently grew rapidly across the US, with email and Usenet newsgroups interfaces with the ARPANET, resulting in increasing cross-traffic between the networks. In February, 1984, Israel became the first international node on the CSNET, soon followed by Korea, Australia, Canada, France, Germany, and Japan.

CSNET played a central role in popularizing the Internet outside the ARPANET, eventually connecting more than 180 institutions and tens of thousands of new users, who in turn went on to further the awareness and spread of the growing network. One of the most important legacies of the CSNET was the introduction of the NSF to the Internet, which led directly to development of the NSFNET.

In 1988, the managing boards of CSNET and BITNET voted to merge, thereby creating a larger network managed by the new Corporation for Research and Educational Networking (CREN).

Resources. The following sites provide additional information about CSNET: