South Africa's only BITNET-Style Network

From ZaInternetHistory

Table of contents


In due course, a number of computers in South Africa were connected either directly or indirectly via the Cyber computer at Rhodes. Without setting out to do it, the group at Rhodes had developed a network that operated in a manner virtually identically to the Bitnet network. The network used IBM's RSCS protocol to exchange email, it was possible to do remote job entry (RJE) from one host to any other host on the network, printing could be shared, files could be transferred, etc.

The earliest archived message that Mike Lawrie can find that flowed across this network is a message dated 8 March 1988 from Vic Shaw logged into the IBM computer at Potch (VSIRSHAW@PUKVM1). However, it is clear from a message dated 12 February 1988 to Vic Shaw from Mike Lawrie, that at that stage email could be exchanged between a computer at the CSIR (CSIRVM) and any of three computers at Rhodes (RURES, RUPLA, RUPHYS).

This section gives some information about that network, which proved to be a stepping stone on the way to the full Internet connection

Link to IBM at Potch

Having cracked enough of the local networking problems at Rhodes so that email could be exchanged between different hosts that were controlled by different authorities, and having got to grips with RSCS protocol that was used by the Bitnet network, it was now time to attempt the next big step. This was to link the Rhodes Cyber to the IBM computer at Potchefstroom University, and thence to Wits, Pretoria and the CSIR in such a way that all of the Rhodes hosts could use this link. The only option open was via dialup networking, using synchronous modems, because the cost of a leased line was prohibitive. The IT Director at Potch, viz Philip Welman, was extremely enthusiastic about such a linkup, and we had his full support and the keen interest of his networking staff (Jan van Rooy, Henry van der Walt and Sakkie Larney).

For the uninitiated, Potch and Rhodes were at the opposite ends of the political spectrum in the country. Rhodes had a very liberal left-wing anti-government reputation, and Potch (to some people, anyway) had a reputation of being somewhat verkramp (loosely, that means "less liberal than others"). There was also a language barrier, which might seem strange to persons not brought up in South Africa, in that Rhodes was an English-language university, and Potch was an Afrikaans-language university. There was also the not-so-minor matter that the name Rhodes, which was taken from Cecil John Rhodes, a British Colonialist of not such good repute, had an unfortunate association with the Jameson raid of the 1890's, and this raid was an attempt to instigate an uprising in the old Boer republic so that Britain could seize the then new and lucrative gold mines of the Witwatersrand. Potchefstroom's forefathers were very much part of that republic. Well, the bottom line is that none of these differences seemed to matter to anyone who was trying to get the network to operate. There is a lesson there somewhere.

A really complicating factor was that Potch and Rhodes were separated by about 1,000 Km (600 miles). The link did not work at first (or second or third or Nth) attempt, and it took several visits by staff from both sites to make some kind of progress.

In the end, the dialup worked. This was at only 2,400 bps, but that did not matter too much. We could exchange email. So what? Well, some of our friends also said "so what, if we need to communicate with someone at Potch we can phone or fax them". There was no quick reply to this, but when these sceptics returned from visits to the USA, they changed their criticism to support.

For the record, the name of the IBM host at Potch was PUKVM1.

That Cable <blush>

In the process of trying to get the dialup link working to Potch, we learnt a very valuable lesson.

Some of us had been into the Potch computer room, and we saw little else but "true blue" IBM ( equipment. Even the modems and even the cables to the modems were made by IBM. Coming from the thin edge of the net at Rhodes, we inevitably made our own cables, and often tweaked them to minimize the number of strands in the cable, thereby reducing the cost. So, we were somewhat bemused by what Potch was doing.

The cable that we used at Rhodes to connect the modem to the Network Processing Unit (NPU) on the Cyber was a "proper" cable made by some cable-maker, and was not the usual home-made cable that we used at Rhodes. After an extended number of unsuccessful attempts to get the modems to lock onto each other and produce a carrier detect signal, and having eliminated all possibilities, including that "proper" cable, of the problem being at the Rhodes end of the link, we were totally convinced that the problem lay at the Potch end. To cut what would be a long story short, we eventually found out that the cable at our end was quite unsuitable for a synchronous modem, and when we put a proper "proper" cable then everything worked. Whether or not the folk at Potch ever discovered our mistake we know not, because they were also making changes at their end of the link.

This was a very valuable lesson to us, viz no matter how certain you are that the problem with a communication link is not at your end, for sure it always is. Beyond question, this lesson helped no end when we had to deal with a network reaching into the USA.

Theorem: The problem is always at your end of the connection.

Corollary: You will always believe with absolute certainty that the problem is at the other end of the connection.

Access to Wits/Pretoria/CSIR

One spin-off of the dialup link to Potch was that there was now communication between staff at Rhodes and the network in the northern region of the country. Mike Lawrie made use of this extensively in the work that ensued as part of the CUP networking committee, and at the same time he exchanged an enormous amount of email with Vic Shaw, then in charge of the CSIR's computer division (and later manager of Uninet). This led to Rhodes acting as a node on the CSIR's national network, thereby gaining more networking experience. More importantly, it built up an ethos of email communication between Vic Shaw and Mike Lawrie, which continued long after Uninet became a formal entity in 1987.

Jessie Gwynne

The person at Joiner Associates who dealt with the sale of JNET to Rhodes was a charming lady called Jessie Gwynne. Mike Lawrie exchanged a large amount of semi-private email with her, and in due course met her on 15th October 1990 at the Educom conference in Atlanta, and visited her and her husband (Bob) at their home in Madison, Wisconsin. This was well before the TCP/IP packets were flowing into South Africa. It brought home to Mike Lawrie, who at that stage did not really need any more convincing, of the power of email communication and the friendships that it generated at the time.