I was looking for an analogy. Not that I misplaced one, you understand, just that I'm an analogy junkie. My fix came in a most unlikely spot; sitting in a dingy waiting room, while my car was being serviced. There, on a coffee table was-are you ready-a wrestling magazine. The cover featured two bovine brawlers in lycra tights, inflicting assorted bodily damage on each other, under the caption of "Main Event."

Not a bad analogy for what's going on in the midrange at IBM. At one time, IBM had a tidy tag-team: the AS/400 and the RS/6000. They displayed many of the same qualities a wrestler might covet: speed, power, and they looked stylish in blue. As grapplers go, they weren't even overweight, although several recent industry reports intimated that the AS/400 was, er, overpriced.

But like any two top competitors, sooner or later they wanted to find out just who was the best. So instead of beating up on the competition, they turned to bouncing each other off the ropes: a struggle consistent with the supposedly new, decentralized, leaner, meaner, competitive IBM. But old top-down habits die hard and, to management, it made little sense to have two competitors from the same locker room wearing each other out. So they assigned them to two different federations: The RS/6000 would wrestle primarily UNIX-class opponents, and the AS/400 would continue to beat up on competitors who previously grappled with its smaller S/3X cousins.

League rules were outlined in an IBM internal memo. It called the AS/400 the "natural follow-on" to the S/3X, and restricted marketing RS/6000 solutions (by default, the unnatural follow-on) "only when an individual [S/3X] customer does not want to migrate to the AS/400." Then, of course, they will market the RS/6000 "aggressively." But not too aggressively. The memo made it clear that IBM will not approve S/3X to RS/6000 conversion firms as business partners. In fact, IBMers were told "not [to] develop [a] proactive S/3X to RS/6000 marketing program." That's what you might call controlled aggression.

Except that, eh, IBM had already announced such a business partner, Universal Software Inc. of Newport Beach, California, which offers software for migrating S/36 applications to the RS/6000. Confused?

For answers, look no further than money. The AS/400 is being accommodated at the expense of the RS/6000 because IBM really can make a ton more money selling the AS/400. Given that product comparisons are tricky and that benchmarks tell a limited tale, still there are some troublesome questions.

While a customer may understand why the unlimited license version of OS/400 with its built-in relational database, communications, transaction processing monitor, security, et al costs eight times as much as frill-free AIX, it is hard to stretch that understanding to the stunning differences in the cost of DASD and memory.

A 64MB memory upgrade on the RS/6000 costs $16,000. The same upgrade on the 9406 AS/400 weighs in at $54,000. And just to confuse things further, the same 64MB memory upgrade costs only $23,760 if your AS/400 happens to be a 9402. For a customer, it's sort of like the experience you have when you go out for breakfast and the menu reads "1 egg = $1.00; 2 eggs = $1.20." You want to tell the waitress, "Give me the second egg."

Similarly, an additional 25GB of disk for the AS/400 costs about $180,000 more than comparable storage on the RS/6000. It will take some elegantly persuasive marketing to convince customers that these are not "apples-to-apples" comparisons. To John and Jane Q. Customer, DASD is DASD and memory is memory no matter what box they call home.

The base hardware price is also telling. The AS/400 E50 requires the buyer to forego a much bigger bag of shells than does a comparable RS/6000 560-$80,000 worth, to be exact.

So why the difference? In spite of the fact that for every complex question there is a simple answer (and that it is almost always wrong), allow me to suggest one: Pricing is directly reflective of the market in which a product competes. Originally, the AS/400 and the RS/6000 were developed to compete in very different and very specific markets: the former, a business system; the latter a high-powered workstation. AS/400 pricing is consistent with historic precedents for business systems. Its success attests to its competitive pricing. The RS/6000, on the other hand, had to break into a much different market which included budget-poor universities and a sophisticated R&D community that was only interested in engine size and unwilling to pay for bells or whistles.

Apparently, no one at IBM quite figured on the crossover potential for these two systems, especially the RS/6000. Sure, AIX is about as friendly as a stopover in Bosnia-Herzegovina, but what if someone were to slap a S/3X shell over it? Someone did. The aforementioned Universal Software offers Open RS/36 which ports application software, standard S/36 programming utilities and files to the RS/6000.

Alternately, the AS/400 wants to play in the open systems arena and is making noises about being POSIX compliant. Both systems are also going after the lucrative mainframe market. It is difficult to disguise something with 24GB of disk and 1GB of memory as a midrange system. So internal memos notwithstanding, the competition is likely to heat up. Customers like Eastman Kodak will see to that. Kodak must have seen some advantages to the "unnatural follow-on" because they are doing exactly what IBM is seeking to discourage: migrating from the S/36 to the RS/6000. And if there is profit to be made in aiding such migrations, firms will not be constrained by the lack of IBM Business Partner status.

Complicating the issue is the question of just who will be holding the tether on the combatants. Both Advanced Workstation Division (AWD), home to the RS/6000, and Application Business Systems (ABS), proud parents of the AS/400, would like to be spun off as independent units, much like the new Personal Computer Company. Rumor has it they share something less than a fraternal fondness. Once free of the squelching influence of MaBM, they will set their own agendas and compete in every market available to them. Therefore, IBM will likely hold on to both until dwindling customer demand-for such items as AS/400 memory and disk-forces prices down to RS/6000 levels. Once profit has been eliminated as a motive for keeping AWD and ABS in-line, they will probably be granted independent status.

Whichever way it goes, IBM is not likely to lose. They have two desirable products. Both will continue to succeed, and both will continue to compete. And that can only mean more choices and lower prices for the customer. Not as exciting as the Hulkster beating up on Junk Yard Dog perhaps, but then what is?