It already begins to look as if the worst-case scenario for the outcome of IBM’s gamble in making the RS/6000 competitive with the leading offerings in the Unix market may be about to unfold – the machine is attractive enough to steal sufficient business from the company’s proprietary base to start seriously hurting the bottom line, but not is not sufficiently successful to sweep all before it and generate the vast sales needed to replace profits foregone on proprietary systems. According to Computer Weekly, IBM UK is doing so badly with the AS/400 that the company received a visitation from on high a couple of weeks ago – John Akers dropped in to ask the reason why. The paper says that IBM UK’s plan called for sales of 3,000 mid-range machines, mainly AS/400s and RS/6000s, this year, but that by the end of June it had sold only 1,000, and found an IBM agent to put the blame firmly on paralysis in the market caused by the RS/6000.

RS thingy

US figures culled from the Computer Intelligence database by Computer Systems News found that in a survey of 6,000 sites, as of April 21, suggests that 30% of AS/400 sales were to new accounts or as additional machines in existing accounts. The proportion converting from System 36 was 21%, down from 27% in April 1989, and only 17% came from System 38 sites against 35% a year ago – but that’s because most System 38 users – the easy AS/400 sales – have moved already. If the profile of new AS/400 sales in the UK is anything like that in the US, the new accounts and System 36 conversions are extremely important to keep the market bubbling – and those are the sites most at risk from RS/6000-induced paralysis. The hype that accompanied the RS/6000 at its launch can hardly have failed to arouse the interest of all IBM mid-range users whose machines were coming to the end of their useful life – and was sufficiently widely broadcast that it will have got to the ears of most chairman. Picture the typical System 36 user. The machine is clearly a dead end but the chap in charge of computing – data processing manager may be too grand a title – has heard enough from his siblings about how hard it is to upgrade to the AS/400, how impossible to do it without putting up the computing budget substantially. But he’s got to have a bigger machine. He’s interested in the touted price-performance of the RS/6000 but is worried about the heat he’ll get from IBM if he says he wants to jump ship and duck the AS/400. He decides it must be an AS/400 Model 25 and tells the chairman so. The chairman says Hey, what about this hot IBM Unix box, the RS thingy? Yes sir, but they can’t deliver the thing yet – they can’t get the software to work properly, and we ought to see it at work at a few other sites before we make a decision. The chairman inwardly hugs the man. In the present economic climate the last thing he wants to do is to burden the company with a big new capital commitment. Well they’ve got it right now, haven’t they – I think we’ll wait till we have some reports back from the field on how it performs before we make a decision – OK? Next business. In the end, the company will probably go for one of the new AS/400 models due to be launched tomorrow, which are designed to ease the migration of System 36 users – but the cost-benefit may still come down on the side of the RS/6000, and almost certainly will in all those virgin target sites for the AS/400.

Paralysis

Meantime IBM hasn’t sold anything where without the RS/6000 paralysis both types of site would almost certainly have committed to the AS/400 by now. But the fact that such uncertainties have been raised in users’ minds is something to give IBM and its shareholders a Merino flock full of sleepless nights. Any AS/400 prospect that opts for the RS/6000 instead is a disaster for IBM. In the first case the margins on the AS/400 hardware are much better; in the second, in any AS/400 sale, IBM gets most of the software revenues – and it has an in-built database where the first thing an RS/6000 sale does is hand business over to Oracle or Ingres;

in the third, a user lost to the RS/6000 is a user lost to Unix. The whole point of open systems and Unix is that users are no longer locked in to one supplier, and there is every chance that IBM will over time alienate RS/6000 users enough to set them running to another Unix vendor when the time comes to upgrade. Alienate? Well it’s already doing its best to alienate its most loyal AS/400 users, the ones that came from System 38 and are already on the largest models and are desperate for the machine that IBM promised would double the performance of the Model 60 within two years when the line was launched, but hasn’t announced. Hasn’t announced because it fears that too many 4381 and small 3090 users would trade down. It knows it can afford to do that because AS/400 Model 70 users have nowhere else to go. But it can’t afford to treat its RS/6000 users with the same contempt – try it and they’ll be straight on the phone to Hewlett or Sun. And in Europe it’s too late for IBM to back-track on Unix. In the US, the company is still promoting the RS/6000 as a technical machine not really suited to commercial applications – but it doesn’t really need to be so emphatic there: most commercial users in the US are still very doubtful about the benefits of Unix. No such reticence in Europe – and IBM did all it could to help the message along. Finding it couldn’t sell the RT as an engineering workstation, but with quotas to meet, IBM Europe went out energetically and promoted it for business in an effort it must be beginning to regret deeply. If an uncompetitive machine like the RT could be sold successfully to European business users, how much more successful should be the vastly better RS box. The demand is clearly there, but IBM has fatally handicapped an RS/6000 that needs a quick take-off: it has said that at some time in the next two years or so it will switch to OSF/1 from AIX.

Hard-bitten

It assures everybody that the switch will cause no disruption to applications, but then it promised AS/400 users double the power within two years – better hold off from commiting to the RS/6000 until OSF/1 is up and running, say hard-bitten software developers, while canny users, convinced by IBM’s promotion of the benefits of Unix, may well decide to go Unix – but not bother to wait for OSF/1 on the RS/6000 but go with somebody else instead. IBM is left with a nightmare to manage – and few are in a more unenviable position than UK managing director Tony Cleaver, who has been given the poisoned chalice of having to promote IBM Unix right across Europe while finding that his home AS/400 base is paralysed and he can’t sell RS/6000s either because Austin goofed on the software. How did IBM get itself into this ghastly situation? By putting short term profits ahead of the needs of its customers and by trying to get the best of both the proprietary and open systems worlds. If it had priced the low-end AS/400s to give System 36 users an affordable performance increase in emulation mode, its AS/400 base would be twice the size it is now and there’d be no question of mass defection to Unix. If it had brought the 4391 out when 4381 users first began to need it, shrugged at small 3090 users trading down, and let AS/400 performance to rip at the top end, the AS/400 would be the fastest-growing base in the industry and bidding fair to become the most profitable – and IBM would have happy customers.