The two manufacturers of IBM- compatible mainframes, Amdahl Corp and Hitachi Ltd, are about to face a grave challenge to their survival. It is not the threat posed by IBM Corp’s CMOS mainframes: they are still hobbled by small engines. It is not the potential of Parallel Sysplex technology: Sysplex compatibility is an engineering problem susceptible to large doses of talent, work and money. It is not the diminishing population of large mainframes: there is still more than enough demand to support three makers of very large commercial systems. If IBM hammers its mainframe rivals during the next year or two, IBM’s lawyers will probably deserve more credit than IBM’s marketing executives. But IBM’s likely victory in its battle against the 1956 anti-trust consent decree will unleash the full power of IBM to compete in the large systems market. Later this year or, more likely, in 1996, IBM may no longer be encumbered by the consent decree. There will be no special restrictions on the company’s activities in the used computer market.
It will be able to offer more packages of equipment, software, services and maintenance. It will no longer be obligated to share interface information with others. IBM’s direct competitors in the MVS market are not without weapons of their own to wield. But, fundamentally, there are only two: the performance of systems and their price. To succeed, Amdahl and Hitachi will have to offer products that are markedly superior to IBM’s in both these aspects. This is possible. But it is also very difficult, and it will become even more daunting if IBM achieves its stated technical goal – delivering a CMOS engine faster than 60 MIPS – before 1997. (IBM is expected to break the 40 MIPS barrier next year and then get past 60 MIPS per engine the year after that). Amdahl and Hitachi are acutely aware that IBM could, before very long, be in a much better position to knock them out of the market. This concern has goaded IBM’s mainframe competitors to action. Each has its own plan. Amdahl will continue to market its current 5995M series of large engine mainframes as long as there is demand for them; in addition, it will deploy the Millennium series of CMOS servers by the middle of next year. The Millennium machines will be nearly twice as fast as IBM’s quickest 9672s and similar in power to IBM’s forthcoming CMOS generation. They will use nine-chip engines that run at 45 MIPS, mounted on a Central Electronic Complex with three to eight engines.
By Hesh Wiener
Total Central Complex power is expected to be 285 MIPS to 315 MIPS. The Millennium processors will use Parallel Sysplex technology to achieve large single image systems – and until and unless Amdahl develops its own Sysplex controller, multi-rack machines will be linked by IBM 9674 switches. Hitachi currently offers its GX 8000 systems, which closely resemble IBM 9021s, and it will ship its first ultra-powerful Skyline processors very soon. Next year, (or possibly in 1997), Hitachi will have CMOS mainframes. Hitachi has licensed IBM System/390 chip sets, so its engine power will be the same as IBM’s. However, Hitachi will build its own Central Electronic Complexes. Despite their ambitious product plans, neither IBM-compatible vendor is yet adequately prepared to combat IBM’s emerging mainframe marketing strategy, which is based on the bundling of processors with other necessary goods and services. The processor component of IBM’s deals is only a fraction of the total cost: system software can exceed the cost of a large processor during a full generation of three to six years. As a result, Amdahl and Hitachi will have to persuade customers that their computers are stunningly superior to IBM’s offerings. Only then can they hope to offset the attractions of IBM’s well-planned total systems package. Further, even if Amdahl and Hitachi can provide machines that are sufficiently fast, cheap and functional to offset IBM’s CMOS and bipolar processors, they may still have to consider another big weapon in IBM’s arsenal
: the Integrated Systems Solutions Co, IBM’s systems integration company. Hitachi has had an excellent opportunity to provide a full range of services through its partnership with 20% shareholder Electronic Data Systems Corp in Hitachi Data Systems. Yet Hitachi Data failed to see the strategic direction in which IBM would lead the large systems business. Hitachi Data should have worked more closely with Electronic Data Systems to reduce users’ reliance on IBM software and thereby offset some of IBM’s ability to control large accounts. Amdahl had the same opportunity to form coalitions with computer leasing companies, with developers of vital mainframe software (such as Computer Associates International Inc and Oracle Corp) and with providers of applications conversion services that could offset some of IBM’s power.
Customers that worry that Amdahl and Hitachi may not have sufficient commitment to the mainframe business should put their fears aside. Fujitsu Ltd (which supplies Amdahl’s mainframe products) and Hitachi each have large home mainframe markets from which to launch big systems and their peripherals for the rest of the world. But neither of these Japanese manufacturers has been able to achieve abroad anywhere near the market shares they enjoy at home. If IBM’s mainframe rivals have the capability and the courage to bring prices down ahead of IBM, they could pick up quite a bit of market share, particularly if they also offer performance advantages. They must begin this process now, well before their new products are generally available, and keep the pressure on as they roll out their next generations. If they hesitate, customers will be inclined to stick with IBM, probably to an even greater extent than in the past. From the September 1995 issue of Infoperspectives International, published by Technology News Ltd, 110 Gloucester Avenue, London NW1 8JA, phone 0171 483 2681, facsimile 0171 483 4541. Copyright (C) 1995 Technology News Ltd. All rights reserved.