On Tuesday last week, IBM made its first serious stab at the distributed processing market with a cornucopia of communicat ions programs. Hesh Wiener argues that the certainty that much of the code won’t work first time, and that it will prove hard for unskilled users to master, gives the first rank players a period of grace to move ahead of IBM once more – but that the future looks bleak for the second tier.
With its 500,000 lines of new code announced last week, IBM has apparently shown most of the cards in its data communications hand. The new hardware that IBM favours for the farflung outposts of interconnected systems is the 9370 described last year, now available on the spot from IBM as well as third party lessors and value-added resellers. The software is that which IBM has just announced, most of it to be distributed during 1988. IBM’s principal rivals for the hearts, minds, sites and budgets of users are clearly DEC and Tandem Computers, followed at some distance by other makers of mid-range systems and fault-tolerant transaction processing computers.
Some industry pundits suggest that the rapid growth of DEC and Tandem has come at the expense of Big Blue. Yet there are also indications that this is not the case. For the most part, users who have installed DEC or Tandem systems to link dispersed offices to central mainframes simply could not even consider IBM’s prior offerings. Business was not taken from IBM. IBM never could have gotten it in the first place. The incompleteness of IBM’s communications software, exploited by competitors’ sales forces for years, was underscored by the magnitude and diversity of last Tuesday’s promises. For IBM’s adversaries, marketing has been a matter of convincing customers that there were substantial, serious companies capable of delivering what IBM did not offer. Now the old sales techniques of non-IBM vendors will lose their effectiveness. IBM still cannot provide complete distributed systems today, any more than it could last week: the general availability of all the programs that users require is months in the future. Nor is it clear that any software as sophisticated as that announced by IBM will work satisfactorily in its initial versions. It’s hard to believe the company’s programmers can deliver everything on time. Software slippage is a problem that plagues all the vendors, not just IBM… and all the users who write applications programs, too. IBM’s internal software development problems are only the beginning of its headache. The new 9370 software, unlike most of IBM’s other 370 family programs, has to be put out in the field for use by non-technical personnel. So it’s got to be a lot better than most of the programs Big Blue fields. There’s no room for software crashes, or even hard-to-use programs. User-written applications face the same challenges. IBM may be getting ahead of itself. Not even the industry giant can provide field support for dozens of new programs running at hundreds of locations… and still turn a profit. User enterprises can’t afford to hold the hands of everyone with a terminal, either. IBM must get each of its customers to take one 9370, load it with untested software, perfect it at the home office and then, finally, replicate it at remote locations. IBM’s announcements – stripped of hoopla – underscore this sensible approach. Yet at the same time, egged on by the IBM promotion machine, observers on and off Wall Street seem to be thinking IBM is capable of working miracles, and overnight, too. The fact is that IBM’s complete renovation of distributed processing software cannot be completed quickly, nor can it be digested by users very soon. This gives other computer companies a chance to re-examine their own offerings. The solutions provided by DEC, Tandem and others may not be as easy to set up as those IBM has promised, but they are at least field-proven. Big companies can buy machines from these vendors knowing that there is a pool of experience. In short, IBM is now the new kid on the communications block, and will b
e for as long as two or three years. There will be a delay before IBM can harvest what it has just sown. Yet the company’s competitors can’t use this pause as a vacation. In order to maintain their momentum, IBM’s adversaries will have to do several things: simplify the installation and hookup of their systems when used in conjunction with IBM mainframes, maintain price-performance leadership, and provide features that remain unavailable from IBM.
Sits on its laurels
None of these things is exceptionally difficult, but all of them require a determination by the vendors to fight ever harder for business. If DEC, Tandem or any of the smaller computer makers sits on its laurels – whether due to arrogance or simply to excessive faith in current products – it will be destroyed. The upshot is likely to be a diverse market, one in which three or four companies can share in the distributed systems business. It doesn’t look as if there’s much hope for any general-purpose minimakers beyond this first tier. Some companies that are presently successful will fall by the wayside if they cannot develop easy-to-use IBM-network-compatible products. Among those at risk are Wang, Prime, Data General, Stratus (even with its sales to IBM), Hewlett-Packard, AT&T and so forth right on down the line… and DEC and Tandem up the line, too, if they slip just once. Makers of conventional and fault-tolerant mincomputers can’t take refuge in the apparent resilience of non-IBM mainframe makers. Users of minicomputers are not so completely locked in with machine-specific software as are those of non-IBM mainframes. On the contrary, they have great freedom of choice. For the most part, the distributed systems jobs that will be set up during the next few years are new applications. IBM is probably shooting for half to three-fourths of a booming market it hasn’t ever really addressed. The non-IBM makers of midrange systems can thrive on much smaller slices of this big pie. But there’s still not enough to go around for all of them.