The 6,000 dollar men
Forget Steve Austin, says Sean Hallahan. When it comes to Unix boxes,IBM's RS/6000 is truly bionic
Big Blue was the last of the major computer companies to acknowledge that there was a demand for Unix machines when in 1990 it launched the RS/6000.
IBM had previously dabbled its toes in the Unix waters when in 1986 it launched the RT PC, a desktop Unix workstation. The machine proved a failure and analysts estimate that it did not even manage to secure two per cent of the Unix market worldwide. The launch of the RS/6000 was IBM's first serious attempt to break into the Unix market.
Even then the company attempted to protect its installed base of mainframes and mid-range proprietary machines by positioning the system as a scientific and engineering workstation. IBM's customers had other ideas and rapidly began demanding that the RS/6000 be kitted out to run commercial software applications.
IBM was forced to comply, although reluctantly. In 1988 it launched the AS/400 as a successor to its earlier range of departmental commercially based minicomputers - the System/36 and System/38. Clearly, the use of the RS/6000 in a business environment could cut across AS/400 sales.
By 1992 other firms such as ICL, Hewlett Packard, Sequent and Pyramid were already heavily backing Unix implementations of business software on their machines. Even Digital, whose founder, and then president, Ken Olsen, was known to be deeply hostile to Unix, was forced to adopt a more conciliatory attitude to the Unix/ open systems market. Today the RS/6000 is an important cannon in IBM's armoury.
Some analysts predict that IBM will eventually merge the AS/400 and RS/6000 ranges into a single machine. In a report published last year - New Computer Hardware: Options and Comparisons, consultancy firm Bloor Research clearly takes this view. It argues that the price structure of the AS/400 puts it at a disadvantage to the RS/6000 and other Unix boxes.
'Ironically the AS/400 was promoted as a Unix basher, and it now looks as though it will gradually take on more and more of the form of the RS/ 6000. Certainly, IBM cannot want to continue development of two such ranges and, if they can modify the AS/400 to the point where its applications can be easily moved on to the RS/ 6000, that would solve the problem,' says the report.
IBM has now thrown another spanner into its own well-oiled works with the release of the Multiprise 2000 low-cost mainframe machine. This runs the proprietary VM/VSE and OS/ 390 operating systems, but with the use of open systems adaptors can appear as a Unix machine. Like the RS/6000 and the AS/400, the Multiprise 2000 looks destined to be sold through the channel.
Robin Bloor, chairman of Bloor Research, no longer believes that the RS/6000 and AS/400 systems will converge. 'I think we were completely wrong about that. There are people in the AS/400 division of IBM who worked on the original development of the machine and are now scattered throughout the corporation. The AS/400 boys are fighting within the corporation as if they were another corporation,' he says.
Bloor questions whether the mid-range RS/6000 boxes have any future at all in the IBM world. 'A lot of people had surrounded their IBM mainframes with HP Unix, Pyramid, Sun, and Sequent machines. The RS/6000 was an attempt by IBM to get its children back. If you can run MVS and Unix on a mainframe why would you need the RS/ 6000?' he says.
Bloor still believes that there will be a role for the high-end RS/ 6000 SP machines for parallel processing and the boxes will still be used by scientific and technical customers.
IBM currently has two distributors, Bytech and CST, which together supply 300 dealers. It also has another 24 resellers with which it deals directly on the grounds that the volume of their business makes it worthwhile.
According to Diane Oubridge, channel development manager for the RS/6000 division of IBM, business has been particularly good. 'Sales of the SP machines are rocketing this year. Not all companies are using the SP as a parallel processing system. Some are using it to distribute applications round the nodes - what we call the Lan in a can,' she says.
Oubridge claims the arrival of the Multiprise 2000 will have little impact on the RS/6000 mainstream market. IBM's Enterprise customers will find it useful when they are bridging between their mainframes and Unix systems but I do not see it has having an effect on the bulk of our customers,' she says.
Although the parallel processing SP systems are doing well, according to Oubridge, and contribute the most in terms of revenue, they are not mass-market machines. IBM claims about 80 are installed in the UK and over 1,000 worldwide. Oubridge says figures from IDC show that the RS/ 6000 is leading the pack in the mid-range server market. While this may be true, IDC still places the RS/6000 division in the overall number three position behind Sun and HP.
In August, IBM responded to criticism that the RS/6000 was out performed by comparable configurations from Sun and HP by upgrading the Power PC chip set from 601 chips to 604 chips. Although benchmark tests proved Sun and HP machines still offered better performance, analysts predicted that by next year there would be scarcely any difference between the rival machines.
All firms have sales peaks and troughs, but IBM's are more unusual than most. The company does half of its business in the fourth quarter of the year, with the remaining half spread over the other three quarters, according to Phil Payne, an analyst with consulting group Sievers. In all probability this is a result of large corporate budgetary years, rather than financial years, running from January to December.
John Saw, technology marketing manager for Hewlett Packard, believes IBM faces considerable difficulty positioning itself in the channel because it is technologically 18 months behind the other major players.
'The channel typically needs to have the best parallel processing because it does not have the reputation of an IBM or HP,' he says.
Saw believes that the larger customers would prefer to deal directly with the manufacturers, but that their channel strategy prevents them from doing so. He argues that the resellers can bolster the confidence of the customer by pointing to the fact that they have the latest state-of-the-art machines which IBM cannot provide.
The hugely profitable RS/ 6000 SP machines are being sold directly by IBM rather than the channel, says Saw. 'They are a direct product because they are hugely complex machines.' He rejects IBM's claims that IDC shows the company leading the mid-range server market. 'IBM did have the right to claim that 18 months ago, but that was not repeated last year,' he says.
Last month both HP and IBM laid out their road maps for 64-bit computing.
IBM predicts that it will be five years before 30 per cent of its customers move to 64-bit computing and 10 years before 60 per cent to 70 per cent take the plunge. Saw concurs with the view that it will be at least the end of the decade before people move to 64-bit computing en masse.
It may be possible to produce 64-bit machines before then, as Digital has proved with Alpha, but until there is a significant base of applications software available the take up will remain slow. Digital's much trumpeted success with the Alpha is based on a relatively small application base in need of high-speed powerful processing.
Martin Finn, marketing director of Hamilton Rentals, a Digital, HP and PC reseller and rental company, believes that while Digital has gone for leading edge technology with Alpha, the RS/6000 group is doing well in the market. 'I have to say that the RS/600 is doing extremely well. Whereas Digital has chased the technology, IBM has been very successful in the past 12 months and recognised the value of the channel which they did not do before,' he says.
This view is shared by Russell Plumber, RS/6000 product manager for CST, who believes IBM has come to understand the channel better in the last year. 'They have been much better at marketing and they have made some very strategic moves within the business market,' he says.
Plumber supports the contention that IBM is highly successful in the commercial sector rather than the scientific and engineering sectors, where the RS/6000 was originally positioned.
AIX, IBM's implementation of Unix, is generally acknowledged to be more secure than other Unix offerings, which makes it an attractive proposition for commercial establishments wanting to run an open systems box.
Plumber believes the large parallel processing SP systems are ideal for applications such as data mining and data warehousing. These are the boxes that IBM either sells directly or through its top-tier resellers. He does not think the bulk of the channel has the skill set to market machines like the SP.
IBM has never been completely happy with Unix. Its heritage lies within the proprietary mainframe and mid-range OS field. The advantage to IBM of the proprietary systems was that the customer was effectively locked in to the environment. Moving to another OS and hardware meant that applications had to be completely rewritten, a costly and time-consuming exercise.
Most customers were prepared to suffer under the rigid regime of IBM and continue to upgrade their existing machines.
The arrival of Unix and open systems changed that perspective. Unix is supposedly a portable operating system which allows users to switch easily on to another platform. Although the ease of portability has been greatly exaggerated by the marketing departments of the Unix vendors, the customer base liked the idea, particularly if it meant that they could get out of paying IBM's software licence charges. In the end even IBM was forced to take Unix seriously.
Today IBM has the RS/6000 positioned alongside the AS/ 400 and the S/390 as a commercial system suitable for business applications. It is likely that the other divisions of IBM would prefer to see the RS/ 6000 squeezed back into the role of a scientific and engineering workstation, but neither the channel nor the customer base will oblige.
IBM is set to make a series of major RS/6000 announcements that will strengthen the machine's role in the commercial environment to become a major force.
The channel, the customers and IBM's rivals now recognise that this force cannot be dismissed lightly.