IBM’s shares may be a buy on a very short- term view, but if the company continues to depend on its top-end mainframes for the overwhelming volume of its profits as it does now, then the shares are very definitely not for widows and orphans to tuck away and forget about in the hope that […]
IBM’s shares may be a buy on a very short- term view, but if the company continues to depend on its top-end mainframes for the overwhelming volume of its profits as it does now, then the shares are very definitely not for widows and orphans to tuck away and forget about in the hope that they will keep them warm in their old age. That seems to be the conclusion to be drawn from Gartner Group’s considered analysis of the company’s September 5 Summit announcement. Katy Ring reports.
Speaking at Boole & Babbage’s Automation 2000 conference in Geneva, Jeff Schulman, vice-president and service director of software management strategies for the Gartner Group gave his opinion on the software arena surrounding IBM’s new and more-or-less new ES/9000 systems. In Schulman’s opinion, IBM showed itself being defensive on hardware vis a vis the plug-compatible manufacturers rather than offensive with its ES/9000 announcement. Therefore, the emphasis is now firmly on software to differentiate IBM’s processors from those of its competitors. Meanwhile software-only architectures will come into their own and the movement towards open systems will accelerate. The intellectual leadership among software vendors will pass to the independents. Despite a shift towards software-only architectures, Schulman detects a sleight of hand on IBM’s part: namely that although MVS/ESA, VM/ESA and VSE/ESA all share the same monikers, and therefore one assumes a level of affiliation, there is very little shared architecture or machine code that is obvious between them. Schulman suspects that this is all a ruse to convince users that high-end IBM software is value-added.
He also commented that new MVS SP 4 machine instructions are designed for their short-term inconvenience to plug-compatible manufacturers rather than for their help to users. Schulman believes that there will be no real software differentiation between the ES/9000 processors and their predecessors until 1993 at the earliest, which will leave IBM vulnerable to the used 3090 market where residual values are dropping, as well as to incursions from more sophisticated plug-compatible hardware. Advanced Program to Program Communications will begin to transform MVS SP 4 into an integrated server so that MVS can talk to Systems Application Architecture machines on a peer-to-peer basis. By 1992 APPC will replace, TSO leaving CICS and IMS as a set of application programming interfaces. Schulman says that Sysplex will improve JES 2’s peer-coupled structure, giving it parity with JES 3 in some areas and surpassing JES 3 in others. Furthermore, Gartner believes that IBM will provide minimal JES 3 support until it is ready to sell the migration from JES 3 to JES 2. Until that time, the Group says IBM will do its best to keep the JES discussion obscure and it recommends that JES 3 customers pursue alternative hardware to that provided by IBM. Schulman says that while collapsing the various versions of VM into VM/ESA makes life easier for IBM in terms of support but does little for most VM users. –
-Indeed, VM/ESA’s dependence on the ES/9000 will restrict user migration – licences will continue to decline and Gartner reckons that by 1993 only 25% of VM users will have migrated to VM/ESA. As for VSE, Schulman says that IBM has a user base of 20,000, which is not moving to other systems. Around 400 users per year are believed to move over to MVS, but as they do so, more come in at the low end. Schulman says that 60% of 370 machines are sold with VSE – nearly twice as many as IBM had thought would be – of which many are new VSE accounts. Consequently, Schulman thinks that IBM will push VSE prices up by around 25% by adding functionality.
Furthermore, in Schulman’s opinion VSE will never be developed to run on anything bigger than a uniprocessor, nor will it ever be compatible with Systems Application Architecture. As for DB2, well as Gartner puts it: eloping with DB2 because your previous DBMS vendor has jilted you is not a successful strategy. Schulman explains that while IBM has managed to sell
5,500 DB2 licences to date, it has been far less successful in bringing DB2 applications into production. Consequently, users struggle to replace their current applications written for IMS or whatever with DB2 calls and thereby create up to a 200% performance penalty for themselves. In order to use DB2 appropriately Schulman believes that customers must rearchitect their systems including the use made of processor, memory and disk subsystems. In short moving to DB2 is no simple conversion.
Gartner says that more thoughtful users are considering whether they can bypass DB2 applications altogether, but use DB2 as a staging area for data to be deployed on cheaper MIPS on local area networks. Moving onto SystemView, Gartner sees this as a bold architecture, with few near-term deliverables, which will limit the development of other systems management offerings for the 370 market. In Gartner Group’s words: SystemView is a significant visionary plan. However, its impact will be limited by its lack of enterprise coverage, its delay in implementation and the non-integratability of the independent software vendor products which will form its core. Indeed, Gartner thinks that the long delay in implementing SystemView will solidify the market position of entrenched, non-integrated vendors, vendors with the largest installed share in their respective niches are likely to be selected as strategic IBM partners. Meanwhile, Schulman believes IBM is storing up further trouble with its SystemView strategy because it is hinging this strategy on the acceptance of OS/2. However, OS/2 is by no means a standard in data centres, whereas Unix is already gaining a lot of ground.