The decision of Michael Blumenthal to cede the reins at Unisys Corp to James Unruh in April makes it time to review his record at the company he created in 1986 with the acquisition by Burroughs Corp of Sperry Corp, and it is clear that, as our headline declared on Monday (CI No 1,352), there remains a whole lot more to do. Unisys has taken the plunge into open systems fairly successfully – but then that strategy was already in place at Sperry when it was acquired in 1986. Indeed it is possible to argue that with the new generation 2200s mainframes shipping in volume and the Unix business, almost all of which stems from the Sperry initiative in 1984, now representing getting on for 10% of turnover, Sperry today would look a whole lot healthier than Unisys as a whole – if it had sold all its defence businesses in the interim. The white hope for Unisys now is its own answer to IBM’s Systems Application Architecture, a set of specifications that will make applications more easily portable between its disparate product lines.
Unmade bed
But that sounds horribly like a beautiful counterpane laid over an unmade bed. Why has DEC been able to soar past Unisys in terms of turnover in the three years since the latter was created? Let us count the reasons. At Unisys, there’s OS1100. There’s OS/3 for the System 80. There’s MCP for the A series mainframes. There’s MCP for the V series mainframes. There’s the operating system on the DCPs. There’s CTOS and its BTOS variant. There’s MS-DOS and OS/2. And there’s Unix. How about DEC? There’s VMS. And there’s Unix. The PDP-11 operating systems went on care-and-maintenance years ago. And the real-time VAXeln is like the yucca in the corner – it gets scarcely any attention from one year’s end to the next but continues to perform its modest function. DEC doesn’t even bother with its own personal computers any more, it just buys them in, and all the attention it gives MS-DOS and OS/2 is directly related to its work on VMS and Unix – and the great beauty of Unix is that the whole world is working on it so that the cost of keeping up with the demands of Unix users is shared among a legion of companies. Yet at Unisys, it appears that not one of the operating systems listed above is on care and maintenance: each one has a painfully small user base – anything under $3,000m a year is painfully small these days – clamouring that it be constantly updated to keep pace with what is available for MVS or OS/400 users.
Pare to the bone
The hope has to be that the new man at the Unisys helm will take a long hard look at the company’s frightening inheritence and pare it to the bone. At most, a year from now, Unisys needs to have only OS1100, top-end MCP, Unix and MS-DOS and OS/2 to support. The DCP programs should be converted to run under Unix on one of the company’s Unix machines. OS/3 and V-series MCP, and CTOS/BTOS must go on care-and-maintenance with a rapid exit to Unix created for all of them. But A-series MCP and OS1100 are one top-end operating system too many, and one of them needs to go within two or three years. The difficulty is that neither is clearly dominant. Probably the least painful solution would be for the company to let one base and business or the other go in a management buyout, because both are painfully wasting assets the mainframe business is growing at only 7% a year now, and slow but relentless attrition continues in the non-IBM mainframe world so that IBM and the IBMulators continue to get more than their share of the business going. There will soon be derivatives of Unix that can handle enormous transaction applications like airline reservations and administration as well as either OS1100 or IBM’s Airline Control Program. NCR Corp has show that by unsentimentally dumping all its proprietary operating systems – oh ITX and VRX do still exist, but NCR would not die if both were to vanish almost overnight – and moving its specialisations over to Unix, a big company can counteract the fact that Unix is a depressingly low margin business: NCR’s distributed banking software and
its retail software all runs on the same hardware and with variants of the same Unix operating software as does its Tower business Unix line, bringing to the company priceless economies of scale. This week, ICL dumps the proprietary retail processors from its own business and that of its Datachecker acquisition and moves over to Unix on the new DRS 6000s.
Appendage
Unix is now at the core of NCR’s product line and strategy and ICL is heading that way – at Unisys it is still an appendage, the company’s new business operating system. The Unisys launch last week of two new V-series machines aimed primarily at the banking market was a depressing demonstration that there is still no-one at the company prepared to bite the bullet and accept the realities of its proclaimed strategies. What is particularly depressing is the way the company keeps clutching at the latest fashionable straws – image processing, systems integration – that everyone else is pursuing, and which will no doubt turn out to be as much chimeras as the fashion of 1986, desk-top publishing, which was going to make everyone’s fortune, but instead made just one company’s fortune, Apple Computer Inc – instead of putting its internal house in order. Unisys today is equivalent to the sum of a Control Data, a Wang, a Concurrent Computer, a Data General, a Nixdorf Computer, a Wyse Technology and the largely unreconstructed Convergent Inc acquisition. All that keeps the ship together is the modest economies of scale of common administration and bulk-buying of common peripherals. IBM has all those advantages in spades, and IBM is in trouble. In the brave new world of the 1990s and open systems, that is not going to be nearly enough.
