The signals coming from different parts of the computer market are so contradictory that it is very difficult to discern whether the industry is in a state of boom or bust, or somewhere in between. The answer appears to be yes to all three. At the top end of the mainframe market, where the IBM […]
The signals coming from different parts of the computer market are so contradictory that it is very difficult to discern whether the industry is in a state of boom or bust, or somewhere in between. The answer appears to be yes to all three. At the top end of the mainframe market, where the IBM Corp standard has the field to itself apart from upgrade business for Unisys Corp and Honeywell Bull Inc, customers seem to be suffering from satiety: they have more hardware and more performance than they know what to do with – and still they are not satisfying the perceived requirements of their user departments. As a result, they are reluctant to follow IBM up the performance curve: the 3090Es were not as successful as IBM expected so the company had to come out with the 3090S – but the problem is so deep seated that it is not at all clear that the S and Enterprise Systems Architecture will be more than a modest commercial success in the immediate future. And where users have to have more raw power, price is the prime consideration, and here, Fujitsu Ltd, with a substantial stake in the company, seems to be looking after Amdahl Corp a lot better than Hitachi Ltd is taking care of National Semiconductor Corp. IBM’s chestnuts But if that is the case, why does Digital Equipment Corp seem to be doing so well with the VAX? The answer seems to be that where large corporate users put in VAXes, they are frequently for completely new ones that they have already seen up elsewhere: instead of overburdening the IBM mainframe even further with yet another application, they put in a VAX and put up a near off-the shelf application they are satisfied can be brought up in a reasonable amount of time – with the promise that VMS won’t suddenly run out of power on them and force them into an XA- or ESA-type conversion. If they are solidly in the IBM camp, why doesn’t the 9370 get a look-in? The perception of the 9370 as a wimp and a failure already runs so deep that it is scarcely considered by large numbers of users, who believe that it is hopelessly underpowered for what it is being asked to do, believe that VM is expensive in staff to run, and whatever the facts, believe that there are insufficient applications to choose from. Will the AS/400 pull IBM’s chestnuts out of the fire in its large accounts? Probably not: far too many data processing managers who have spent their working lives talking with sales survivors from IBM’s old Data Processing Division have over the years been so inculcated with the idea that the General Systems Division products – Systems 32, 32, 36, 38, are not for them that if their salesman now tries to persuade them that all is changed with the AS/400, the credibility gap will prove too wide to bridge. Which is not to say that the AS/400 won’t be a success in its own constituency: it looks by far the best new product from IBM in years, and in contrast to the 9370 has the aura of a successful product. But the task of upgrading hundreds of thousands of 34 and 36 sites mean that most of the first two or three years’ revenue will perforce come from the existing base, and the AS/400 is unlikely to give the VAX too much trouble in IBM’s fortress sites. The other problem the AS/400 faces is that the battle for the hearts and minds of hundreds of thousands of new companies that are ready to graduate from personal computers to their first real computer seems to have have been decisively won by the Unix fraternity. Siemens AG’ success with its Sinix machines in West Germany certainly seems to point that way, and with the likes of Unisys and NCR Corp deriving a substantial part of their sales from Unix systems these days it is clear that the commercial vogue for Unix is still in its infancy and has enormous potential – hence IBM’s involvement in the Open Software Foundation.
Workstation boom The other low-end market that is booming is the engineering workstation business: tens of thousands of smoke-stack companies have died over the past decade in the US and Europe, hundreds of thousands more have been forced to become much more e
fficient to survive, and they have been persuaded that the first step to that efficiency is to replace manual design methods with computers, not only to design new products but to compete with the Japanese by adapting those new products to create variations that can be marketed as improvements: few Japanese consumer products have a life of more than a year or so before they are replaced by superficially new models, and to survive, US and European manufacturers have to try to keep up: engineering workstations make that a little easier, and if you buy an engineering workstation, you can have any operating system as long as it’s Unix. That is the reason that Data General Corp and Prime Computer Inc are having such a lean time: based overwhelmingly on off-the-shelf microprocessors, Unix workstations are intrinsically cheaper than the old-style minicomputers running proprietary operating systems can be, and for users wanting something bigger, there’s a wide choice of minisupercomputers which also run Unix. Prime and Data General were each critically late into Unix, and in trying to compete with DEC in traditional minicomputer markets are now facing the brick wall that faced the Bunch mainframers in the 1970s: these days, nobody ever gets fired for buying DEC. Sun Microsystems Inc, virtually a one product company – a product characterised by the combination of Unix with competitive price-performance, looks likely to pass not only Data General but even the Prime-Computervision combination in terms of sales, and the latter two seem doomed to be relegated to niche markets where they can demonstrate clear superiority in the same way that Gould Computer Systems Inc and Concurrent Computer Corp have been.
Meltdown Monday The one vertical market that – pace Meltdown Monday – still seems to be running at full throttle is the financial services sector, but here too, the battle seems to have been largely lost and won, the victors being DEC and the two dominant exponents of high transaction rates coupled with fault-tolerance – Tandem Computers Inc and Stratus Computer Inc. But even that market is showing signs of reaching its peak, and competition from high transaction and sometimes fault-tolerant Unix systems is likely to make it a lot less profitable. And once again the upgrade from personal computers is not going IBM’s way: as like as not, where last year’s brokers’ desks were littered with IBM or compatible personal computers, this year, they are as like as not interspersed with Sun or Apollo Computer Inc or Hewlett-Packard Co Unix workstations. As for the personal computer business itself, Apple Computer Inc appears to have decisively won the hearts and minds of the graphic arts fraternity, which makes the battle to get OS/2 and Presentation Manager established look like a tough one. Where price is not an issue, the buyers of perceived quality buy IBM or Compaq Computer Corp if they don’t buy Apple, otherwise the market has become in John Akers words exclusively commodity-like, which means that most business goes to the recognised name with the best price, which bodes well for Amstrad Plc’s 80386 machines, even if, as seems the case the market is beginning to slow.