Computergram has been preaching the invitability that the Unix operating system would come to dominate the computer industry ever since its launch some five years ago. At that time there was a lively chorus of naysayers demanding why we kept harping on this dreary tecchies’ operating system that was simply incapable of meeting the needs of business users. Our argument then echoed the response of Mae West to the girl who exclaimed Goodness, what beautiful diamonds: Goodness had nothing to do with it. Tecchies’ toy It didn’t matter how inadequate Unix was for the purposes for which it was being promoted: the whole history of the computer industry points in one direction: the systems solution that is objectively the best is never the most commercially successful. It would be hard to find anyone to champion Cobol as the best business programming language, but that did nothing to prevent Cobol becoming by far the most widely used. CP/M was even more of a techhies’ toy than Unix, yet it dominated the personal computer market until MS-DOS, which had been deliberately modelled on CP/M, swept it peremptorily aside. IBM’s 370 architecture is emphatically a batch architecture yet vastly more on-line transactions are processed on it every day than on any other architecture. Even those who have grown to love WordStar would find it difficult to affirm, hand on heart, that it is the best word processing program. And conversely, by definition, OS/2, having been designed from scratch to be the ideal high performance single-user business desktop operating system is now quite clearly destined to be squeezed between MS-DOS and Unix, and within five years is likely to be little more than a footnote in the history books – perhaps as the Ford Edsel of the computer industry. Also going against it is the hard lesson that microcomputer history teaches – that the same company or companies never dominate successive generations, and IBM and Microsoft had their day with the 16-bit generation. Before too long, Microsoft is likely to find that the investment demanded by OS/2 is quite simply not justified by the return it is getting, and as users start to clamour for MS-DOS 5 and MS-DOS 6, it will have to swallow hard, divert its resources into meeting that demand, and turn OS/2 over to IBM lock, stock and development team. Will IBM really be able to achieve more with it than it did with DPPX? Ugly duckling And so it is that what we tagged five years ago as the irresistable tide running for Unix has performed the same ugly duckling into reasonably convincing swan transformation on it that IBM performed on its batch architecture to tack on high volume transaction processing. Today those who decry Unix as unsuitable for general business applications find themselves voices crying in the wilderness, and with the advent of a choice of three or four gonzo-friendly user interfaces for Unix, plus copious solutions for running MS-DOS applications under Unix, just about everything needed for the final push onward to final victory is now in place. IBM will still tell you that Unix is only for technical applications, which is becoming a little embarrassing because in Europe, where the RT has been modestly successful, it has been sold almost exclusively for business applications. And IBM’s plaintive cry points up a trend that has so far gone all but unnoticed: almost all major new computing concepts and architectures have started out in the scientific and technical domain, and have then spilled over into the commercial market. In the early days of computing, the vast majority of machines were used for scientific work, the DEC PDP-11 and VAX-11 both started out as pure scientific and technical machines, and Unix workstations are in process of making the cross-over from CAD/CAM into the commercial world that will soon become their primary raison d’etre. So it is that the embryonic generation of parallel supercomputers, which almost without exception runs some variant of Unix, is already beginning to make the transition from the scientific into the commercial world
. Dow Jones & Co is using parallel processors in super-high volume database information retrieval work, and just the other day, Metier Management Systems, whose Artemis large scale project management software runs almost exclusively on IBM mainframes (and latterly personal computers) announced that it was doing a version for Bolt Beranek & Newman Inc’s new TC2000 88000 RISC-based parallel processor, which runs under a variant of Unix (CI No 1,223). In the supercomputer world, Cray Research Inc is moving over to Unix, and almost all the companies that have come after it have taken Unix as their starting point. The new scientific systems manufacturers that survive will be the ones that are able to diversify their machines into the commercial market, which means that the traditional mainframe with its proprietary operating system will begin to find itself squeezed by Unix from the top down as well as the bottom up. Disaster And all this has to be very good news for users, because once industry standards are firmly established and a multitude of companies is essentially selling variants of the same products, the user is free to pay the price he can afford for what he wants. And a mistake is no longer a disaster locking the user in to an unsatisfactory supplier. Here at Apt Data Services there are two IBM machines, one a gift from a former business partner, the other in settlement of a debt, and three MS-DOS machines from Dell Computer Corp, plus ones from Tandon Computers, Zenith Data Systems, Toshiba Corp, and a no-name Taiwanese clone. Each was the best we could afford at the time we bought it, offering the functionality we needed and support and reliability we could live with. Had IBM had the MS-DOS market to itself, prices would have been vastly higher, and small, fast growing companies would either have been strapped for computer power or would find themselves with plethora of different types of machine, most incompatible with each other. And mainstream users who get onto the Unix curve now will soon be in a similarly happy position of being able to get what they need and name their own price.
