When IBM finally gets to the Summit, as its next generation of mainframes has been dubbed by the press, it may find Amdahl there asking, What took you guys so long? Not only has Amdahl announced that its fastest general-purpose systems will be built of uniprocessors with twice the horsepower of IBM’s 3090-180E engine, but the company also beat Big Blue with a large disk subsystem based on an apparently much more powerful controller and clustered-module drives. To a large extent, Amdahl’s May 3 announcements will provide, several months from now, much of what users expect IBM’s 3090F and Summit systems to offer over the next few years. This puts IBM’s mainframe business in an uncomfortable position. Big Blue’s awkward stance will be familiar, however, to anyone who has paid attention to developments at the other extreme of IBM’s business, the personal computer trade. To the extent that events in the PC business have analogies in the mainframe world, Amdahl’s announcements may presage a significant loss of market share by IBM in the mainframe market that it so cherishes. While IBM’s PC business has been on the rebound during the past year as PS/2 systems gain popularity, at its nadir, the industry giant saw its US market share considerably eroded. By some estimates, IBM’s PCs garnered more than 60% of the business market at their peak, and as little as 30% of the same segment during IBM’s darkest time, a year or so ago. What brought IBM’s PC business new vitality was not so much the new products in their prettier cases. Nor was it the microchannel, high technology’s answer to the pineal gland. Rather, it was IBM’s giving the customer better products and better value, too. Most visibly, the VGA helped IBM win business, as did disk drives that are as competitive as those on the original XT and AT. The mysterious OS/2, while still as bare of applications as Wall Street is of gentlemen, may have played a role, too. IBM’s mainframes do not provide good value for money. Quite the opposite, as IBM seems to be constantly fiddling with the contraptions – at the user’s expense, of course. The new versions of 3090 mainframes compel the user to keep up by buying hardware. In addition, users must tie up scarce human resources reorganising their systems software and applications workloads; this is more costly and confusing than changing hardware. Since their first appearance on users’ computer room floors only two and a half years ago, 3090s have been through a major renovation – the E upgrade – and a minor one – the ESA variation. Another few tweaks – collectively, the so-called F modification – are expected in time for the Christmas selling season, which begins about the time IBM has to show its third quarter results to shareholders. Unworthy Amdahl’s competitive answer to ESA is a system that ought to survive for two or three years as originally announced, with the possible exception of some microcode changes. If IBM’s next few mainframe facelifts don’t involve architectural changes, the latest Amdahl offerings could last even longer. This not only makes Amdahl machinery appealing in its own right, but it also gives the user a stable environment in which to develop new applications. The combination of ESA and the revised DB2 alone will be more than enough for large shops to cope with during the next few years. With SAA added to the stew, few organisations will be in a position to digest large projects easily. IBM’s treating mainframe customers as if their mainframes were fashion items – in this season, out the next – is behaviour unworthy of either party. In a roundabout way, the Amdahl challenge may be good for IBM and all the organisations wedded to IBM’s de facto mainframe standards. IBM may have to cancel some of the modest and confusing engineering changes it is believed to have planned for the 3090. Instead, IBM may have to gather its wits and wiles in order to make a large leap forward, one that would carry its technology past the new boundaries Amdahl has defined. IBM may have to scratch the F kicker and instead plod on tow
ards Summit. Currently, if IBM must look backwards in order to regain its perspective, the risks of fright are at least lower. There is now one fewer company gaining on it. – Hesh Wiener
