TCP.IP at Rhodes University
From ZaInternetHistory
- The title of this article is incorrect due to technical limitations. The correct title is Editing TCP/IP at Rhodes University.
| Table of contents |
Overview
The Internet uses the TCP/IP protocols, so it is relevant to attempt to shed some light on how these protocols were introduced at Rhodes. There was no strong motivation, because the network that evolved worked well enough over twisted pairs of copper wires. All that was needed to connect a PC to the network was that the simple and low-cost cabling should extend to the location of that PC, and that the PC should have a serial card. Initially, this was fairly simple, but the policy of allowing (indeed, forcing) each department to find the funds from wherever for purchasing the PCs that they needed (as opposed to having a limited central pool of such funds) meant that expansion was rapid, and wiring up departments even with simple wiring became a major exercise.
Also, after a disastrous attempt to use Microsoft's LAN manager in the PC labs, we switched to Novell and then the installation of a campus-wide LAN became essential. Such a LAN was an ideal test-bed for experimenting with TCP/IP.
PCIP
The first version of TCP/IP to be used was that of the PCIP package that was developed at MIT and put on public release. In this particular case, a complete set of disks was purchased from Austin Code Works (??date??). This package supported only a few types of LAN cards, and so one of the specific cards was purchased at a wicked price. An IP number was obtained (192.42.99.0) and an attempt was made to try to get something to run.
FTP Software
The natural evolution from PCIP was to move (??when??) to the successor commercial product, viz PC/TCP, that was developed by FTP Software in Wakefield MA in the USA. Rhodes University's political credentials were important in securing this purchase because all too many US-based firms would have nothing to do with any institution based in South Africa. This package provided plenty of flexibility, and we updated things when the packet driver technology emerged. We also purchased the LanWatch package so that we could monitor ethernet traffic - this package alone played an important role in that it allowed us to get to grips with TCP/IP as low as at the bit-level when necessary.
SCO Xenix
Rhodes had acquired a 386 computer with the SCO Xenix operating system some time in 1989. This was a donation arranged by Dave Barnard of Central Data Systems. This acted as a vehicle for learning about the Unix-style of operating systems. This computer was initially given the name RUCCX1, and subsequently when we decided to use an "African animal" naming theme, was renamed QUAGGA. This Xenix host had been augmented by a second one by October 1989. CDCNet
The NPU of the Cyber had been augmented with a CDCNet box. This had some simple TCP/IP facilities, and allowed telnet and FTP. It needed a hostname table file to drive it. By the 22nd May 1989 there were some 5 entries in the Rhodes' hostname table, and this had grown to 13 entries by the 3rd of October 1989.
IP Number Assigned
Rhodes received its first IP network number from the Internic on the 16th September 1988. This was the number 192.42.99.0, and was assigned by Sue Romano of SRI, (whom the Mike Lawrie was to meet in October 1990 in order to discuss .ZA registration issues.)
NCSA and CUTCP
Helpful and supportive as FTP Software was, and as good as their software was, it was simply going to be beyond Rhodes's budget to install this package on each and every PC on campus. In due course, we discovered the NCSA and CUTCP public domain packages. These were fairly good, but were buggy. Some work was done at the University of Port Elizabeth by Darryl Anderson and/or Peter Lock that removed a timeout bug in the underlying software, and this helped a great deal to improve the performance. This fix was to the CUTCP package, but apparently the original authors took little interest in this fix - maybe it simply was too late in the day, and Windows™ packages were becoming the rage at the time.
TCP/IP Workshops
A number of Internet workshops were run at Rhodes University. These were specifically to deal with issues of internetworking, and it was required that the attendees should have some knowledge of how to install and run a TCP/IP package on a LAN.
The first workshop was held on 6th to 8th August 1990. This was specifically to plan a TCP/IP link between Rhodes and Potch, but the concept expanded to become a general sharing and pooling of TCP/IP knowledge and experience.
Attendees at this workshop were Alan Barrett (UND), Angus Hay (Wits), Francois Jacot-Guillarmod (RU), Gavin Melville (UPE), Chris Pinkham (UCT), Jorge Rabacal (CACDS/CSIR), Henry van der Walt (Potch), Jan van Rooy (Potch), Hennie Smit (Medunsa), Mark Stobbs (UFH) and Dave Wilson (RU). Of course, the Mike Lawrie attended as well. The idea behind the workshop was literally to bootstrap the efforts to get TCP/IP connections running on some of the Uninet circuits. By that stage, the TCP/IP links to UCT and UND had been operating for two months, using the PCRoute package, and there was a growing use of this small internet.
More Workshops
There was little doubt that a core of Internet knowledge had built up at Rhodes, and there was also little doubt that there was considerable interest in this topic from various persons in the country. So, we (Francois Jacot-Guillarmod, Dave Wilson and Mike Lawrie) decided that it would be a good idea to share this knowledge, and to get some return on our efforts. Workshop #2 was hardly advertised, word was put out via the network as it then was towards the end of 1990. In no time, the workshop was oversubscribed. The names in the Mike Lawrie's diary for that workshop are, in no particular order, Paul Nash, Paul Ducklin, Mark Elkins, Hans van Staden, Dennis Cronje, Sean Naude, Johan Strydom, Tony Putman, Heidi de Wet, Kuyper Hoffman, David Lurie, Aris Stathakis, Vic Shaw and Hennie Rautenbach. It is not known at this point in time whether these all attended, or whether others attended in their place. This course ran from 7th to 9th January 1991.
Because this second workshop was oversubscribed, we organised a third one. This ran from 29th to 31st January 1991. Regrettably, the the names of the attendees are unknown (??help will be appreciated??).
Value of a boozy lunch
The last day of these workshops included a lunch at a local restaurant, which was but a short walk from the Rhodes campus. Most South African universities seem to have a reputation for alcohol consumption, and Rhodes is no exception. The "last day lunch" of these workshops rapidly degenerated, and no formal coursework was done on the last afternoon. Yet, somehow, it seemed to us (ok, we were looking through the haze) that somehow more was learnt about networking as a result of that lunch. Perhaps it was because the folk on these courses were indeed the real networkers of the country, who were going to make things happen, and they got to know each other so much better. Who knows.
Theorem: There is a lot about networking that is never taught in any Computer Science course.