The Meridian Leasing-backed debate on IBM’s mid-range strategy last week added little to the sum of knowledge known about IBM’s plans. Various IBM experts beat their chests and claimed that their products were best placed to take advantage of IBM’s mid range strategy and even the odd truly independent voice couldn’t point to the strategy that would redeem IBM, giving more weight to the idea that IBM has no way out of its mid-range dilemma. There is still the feeling that IBM will somehow find a way into the hearts and minds of divisional computing and medium sized users, but no-one seemed to be in a position to pull IBM’s rabbit out of a hat for it. Questions such as, Will there be an upgrade path between the System 36/38 camp which has a cast of some 200,000, and the big money spinning environments of the System 370? were asked, and various versions of yes and no were put forward, but no-one could say HOW IBM could make such a move attractive to the junior camp of System 36 and 38 users. Fired In the background lies the imminent announcement by IBM of its Silverlake mid-range machine which the company hopes will claw back the market share it has lost to all the rival minimakers led by DEC. But while speculation was rampant, none of the distinguished international panel of IBM watchers invited to speak at the day-long seminar was able to offer rock solid information on Silverlake, and there are suggestions that it has been delayed substantially. But for those new to IBM’s mid-range sectors it was an eye-opening event, with good summaries of where the various IBM offerings have taken us to date. On the face of it, IBM appears to have settled on a three-pronged assault on the mid-range market – the Systems 36 and 38 drive for the small systems commercial sector; the use of RISC-based machines aimed at the engineering environment and the invention of a baby 370, the 9370, to handle distributed computing. But is this enough to safeguard IBM’s future in the mid-range market? IBM watcher and relational database guru Shaku Atre said that IBM had lost the ‘mind market’ in the small business sector. She was referring to the death of the old axiom Nobody ever got fired for buying IBM, which in the modern world is being added to with but you might get fired if you don’t consider the DEC alternative. Ms Atre said that studies of user satisfaction showed that DEC and Hewlett-Packard were ranked first and second and trailed badly behind. She said IBM’s problem is that it has retained a mainframe mentality to computing and does not understand mid-range and micro users. If IBM does not clean up its act in this area, it’s going to be in trouble, she said. Anyone reading the papers since the launch of the 9370, or since the summer of 1985 when IBM’s mid-range began bearing the brunt of the blame for IBM’s dreadful figures can hardly be in any doubt that IBM already is in deep trouble. Inability This view was echoed by Mike Newman, managing director of Pacific Associates and long time System 38 fan. IBM is in danger of being wrong-footed in the mid-range arena as happened in the PC market. The major reason for the company’s inability to respond to the challenge of the minimakers arises from what is also one of IBM’s strengths – its strong corporate management, he declared. The company remains steeped in the philosopy of the central site and has tended to assume that all the mid range user requires is a System 370 grown small. Yet, as Newman agreed, IBM seems to have realised the shortcomings of this approach and is now revamping its strategy: the drop in mainframe revenue and now PCs sales too, as users try to work out which way to go with the advent of PS/2 – has substantially slowed profit growth and forced the company to cast a long hard look over its approach to the mid-range arena. The new emphasis on this sector is clear. The mid-range management structure has been revamped, with Stephen Schwartz appointed worldwide general manager for mid-range systems – and he is on record as saying that by 1992 3m organisations wor
ldwide will have installed mid-range systems. If even half of them were IBM systems, then IBM might quickly forget its beleaguered mid-1980s. In the UK a director has been appointed to head a new division supporting mid-range marketing. And to qualify for the 100% Club IBM sales staff must sell a specified amount of mid-range systems – back to the old rules of General Systems Division. But IBM is still grappling with a major problem – how to rationalise the mid-range product lines. After all, who wants to be offering a three-pronged attack, when there is only enough cash coming in to support one prong? DEC’s success in the mini market has been spurred by just that, the homogeneity and consistency of its VAX/VMS range. The single architecture route taken by other mid-range manufacturers contrasts sharply with IBM’s mish-mash of under-powered 370s, 36s and 38s soon to be merged into Silverlake, Series 1, System 88 fault-tolerant machines from Stratus – and the introduction of RISC-based machines and Unix to take the place of the 8100 in adding another layer of incompatibility. The majority agreed that the choice for IBM seems simple enough: continue developing 370, Silverlake and RISC separately, or abandon either 9370 or Silverlake, or unify the two under one operating environment. One clue that was thrown into the equation by Bob Trasker of the Yankee Group was that he believed that the undelying architecture of the Silverlake was a RISC. This suggests that in one fell swoop IBM will manage to merge both System 36 and 38 environments, while laying the ground for building an architectural relationship with the RT range of workstations and provide a platform that could also quite easily emulate 370 code. Throwaway Trasker believes that Silverlake is the centrepiece of IBM’s future strategy, the machine has been seen in beta test and is definitely a RISC architecture with some of the ‘best attributes’ of Systems 36 and 38. If IBM gets its marketing strength behind the new vehicle it could manage to reduce the architectures it supports down to two camps only rather than the three-and-a-half at the moment. The final stage of the plan would be to produce a development link between 370 and Silverlake at some stage in the future. But the few speakers who subscribed to this theory had little to offer in the way this could be achieved. Throw-away lines about making Silverlake the back-end processor to the mainframe or that IBM has been working on a System 38-to-DB2 automatic conversion utility, but is waiting for DB2 to go faster are suggestions that have been heard over the past few years, but weren’t taken up by any of the speakers. Newman felt IBM has to throw away all of the 370 architecture one day because it’s so primitive, but even he didn’t believe IBM would actually do it, just that it should. System 370 architecture is simply too complex and intrinsically too primitive to support the burgeoning software requirements coupled with ease of use demanded of today’s commercial systems, Newman concluded.