Overview
From ZaInternetHistory
This document sets out to describe how the Internet was started in South Africa. It is a ramble around some of the background events that caused this to happen. There might be little apparent logic, nor any good sequence, in this document - perhaps time will permit some tidying up to be done. So be it, but everything described in this text had an influence on the startup of the Internet in South Africa.
In brief, a group of three persons, viz Francois Jacot-Guillarmod, Dave Wilson and the author, Mike Lawrie, found a way to establish a sustainable email link to the Internet in 1988. A number of others helped with this exercise, particularly Randy Bush and Pat Terry. The link was established between Rhodes University, in a small city called Grahamstown, and the home of Randy Bush in Portland, Oregon. The email link went into production for general campus at Rhodes in February 1989, although email was flowing across at least two Fidonet links in 1988. Mike Lawrie was the Director, Computing Services at Rhodes University, and Jacot and Dave were Systems Programmers in the Computing Centre. Pat Terry was the head of the Computer Science department, and the author is not too sure of a correct description for Randy Bush, but we all knew that, while he had not the slightest obligation to help any of us, he was a person who gave willingly of his time and shared his experiences. At the time that he started to help us, he was a compiler writer for Oregon Software.
The email link used the Fidonet mailing system as a transport mechanism to exchange email between the Control Data Cyber computer at Rhodes University and a Fidonet gateway run by Randy Bush of Portland, Oregon, in the USA. The Fidonet system in the USA had a gateway into the Internet, and to many people's amazement, including the group at Rhodes, the system worked and stayed working.
One thing led to another, and about a year later there was a UUCP gateway in parallel to the Fidonet link. This was also a dialup system. The UUCP gateway provided vastly superior facilities to Fidonet, because the standards used for email on the UUCP transport protocol were in line with the Internet standards as described in RFC 822. At that time (circa 1989/1990), the TCP/IP protocols were in use internally at a number of South African universities, in particular at Rhodes University, the University of Cape Town (UCT) and the University of Natal, Durban (UND). In that era, use of a 9600 bps channel was obtained via the fledgling Uninet network to connect Rhodes and UCT, and an attempt was made to connect up a small internet between the LANs at these two institutions - this was in (?March 1990). Routers were to all intents and purposes unobtainable, so we tried to use Win/Route - a commercial product sold by Wollongong that was supposed to make a PC act like a router. This simply did not work. Having wasted good money on that product, and about to give up, we then tried the public domain PCRoute, and then traffic started to flow. Work on getting email to flow across this link took place in May 1990. The technical work at the UCT end was done by Chris Pinkham, support at a policy and strategic level at UCT was from Fred Goldstein.
With the success of that internet link between Rhodes and UCT, a second internet link was established to UND. The technical work at that end was done by Alan Barrett. Note the lower-case internet, this was not part of the world-wide Internet yet.
In due course, the UUCP dialup link was replaced with a full Internet connection that operated across a leased line (to Randy Bush) at 9,600 bps. This was towards the end of 1991. For whatever reason, we used the KA9Q package to handle the routing at each end of the circuit - quite why we did not use PCRoute consistently, or KA9Q consistently, has not yet been explained to Mike Lawrie by Jacot or Dave or Randy. It matters not, the packets flowed.
When sanctions against South Africa were lifted by the USA in 1992, a pair of Cisco (http://www.cisco.com/) routers was used on the link to the USA. In due course, Ciscos (http://www.cisco.com/) replaced the PCRouters which at that stage were connecting about 15-odd Uninet sites together.
Now this is all pretty simple to tell in a summary such as this. But that does not tell the entire story. There were many things that had to happen in order to get things to work, and to keep them working. This is the story about those things. Very few of them were planned or even thought about, things had happened that contributed to the success, and the co-incidences are quite surprising.