IBM
0255 System/38
The most important new product GSD Marketing Group were all involved in launching, where which System/34 was something of a diversion, was System/38; code-named ‘Pacific’. All IBM new products were code-named to keep them secret.
It looked from the number alone (S/38) as if it was no great advance on the System/34. In fact nothing could have been further from the truth; it was genuinely revolutionary!
I got heavily involved with this product, along with my colleagues, something like a year in advance the launch. I, in particular, was fascinated by its architectural specification. It was a wonderful machine under the covers. It was something like a generation ahead of its time. Not least it had 48k addressing. This didn't become more generally available for another 20 years. It also had relative-addressing for everything. This was most obvious in terms of the storage. Before this time data stored on disk had been stored in physically separate files. Thus, when you added to a file you wrote in the sector dedicated to this or – if this was full - started a new sector somewhere else. In the case of Pacific every bit of data, and indeed every element within each piece of data, was stored separately and could be put anywhere on the system. There was no such thing as a physically continuous file. Only the operating system could tell you where the data went. What was more, the machine itself was relative. Thus, in theory, it was possible to start processing an instruction in London, continue it a fraction of second later in on a machine in Los Angeles and finish it on one in Tokyo. You wouldn't notice any difference whatsoever, the machine kept track of this. I don't think this was ever implemented, but it was indicative of the incredible power of the machine
This was one very powerful operating system. The operating system is what makes computers powerful; and this one had everything that every system designer might have wanted. I fell in love with it at first sight.
As it turned out, it was a little too far ahead of its time for the hardware. Eventually, a decade, later it ran very powerfully indeed; as the hardware caught up with it. But when it was first launched it struggled. At least the hardware available at the time struggled to run the overhead of the software. There was, indeed, an enormous software overhead. If I remember correctly, it was something like 60 MB for the main programme; which was truly enormous for the time. Now it is quite normal for ordinary PCs to have Microsoft operating systems running into hundreds of MB. But it must be remembered that the ordinary PCs now can have something like a gigabyte of a memory when, at the time System/38 came out, even the biggest mainframes were likely to have as few as 100 kilobytes of main memory.
As usual with IBM there was a master plan for a whole range of machines. There were very small machines to take over from the System/32 and System/34, and very large machines to handle DP division applications. As was also usual with IBM, most of these machines were never actually launched. Certainly the smaller machines were never launched, since the hardware had enough problems keeping up with the software even on the medium-sized machines. This was often the case in IBM. Its original S/360 machine was supposed to have one operating system, OS, but a second had to be developed as a matter of urgency; the new DOS system just for the smaller machines. Also typical for IBM, the two software operating systems were incompatible for the whole lives of the machines in the range. At the top end, on the other hand, it is quite possible that large S/38 machines would have been viable. But they would have competed with the very profitable large mainframes of DP division, and IBM saw no point in cannibalising this market. However, a couple of decades later, the DP division accounts did start buying large numbers of networked machines based on the S/38 architecture.
Not merely was it my favourite, but it was Keith's chance to make his reputation. Accordingly, everyone was directed to work on this new machine. Even though I had the launch of a new machine in the meantime, I had to keep this low-key so that none of the attention was taken away from Pacific (S/38).
There was a significant amount of paperwork that followed the new machine around. Much of this was in the form of what IBM called ‘candy stripes’. These were top secret documents, which were printed on paper that was literally candy striped, in pink, across each page. Within each page there was also not just the page number but also the number of the recipient. This meant that any leaks could immediately be traced back to the recipient; since the numbers were not small, but were the size of the whole page - in pink. I got into trouble when I later left the department, for I handed my candy stripes – of which by then I had a number - to the person who followed me. This was a breach of rules, for I forgot that I had to hand them back to the security coordinator who would cross them off against my name and reissue them to the new manager. What was worse, the recipient had then proceeded to lose them! This was a fireable offence in IBM, so seriously was confidentiality taken. There was a massive internal inquiry, and I only just escaped with my job.
The key documents, which had peculiar names that were only meaningful within the IBM new products operation, were actually intended to obtain country input to the pricing policy. IBM, give its due, took the right marketing stance; of not just adding on to the cost but of actually deciding what the market would bear. Indeed it had a process where it charged high prices at the launch but, through the product’s life, gradually decreased these until - at the end of its life – the stocks left were almost remaindered.
hits