Flashback: Clearly A Mainframe, Nearly The Perfect One

March 26, 2012 Timothy Prickett Morgan

TFH

Back in September 1992, when the E95 quad-processor AS/400 came out and IBM was in the midst of a major reorganization as its mainframe business was imploding and Lou Gerstner was about six months away from being named the first outside CEO of Big Blue in its history, I jotted down a wish list of the things that I thought IBM needed to do to make the AS/400 a more general purpose platform.

Back then, The Four Hundred was only available as a paper newsletter on a subscription basis, so very few of you have ever read this wish list I put together. But I thought, given that we have another new CEO, Ginni Rometty, in charge at IBM, it would be a hoot to drag out the old wish list, which might have been the very first one I ever felt comfortable putting together because I actually knew a thing or two after a few years of covering the systems racket at that point. Anyway, here it is. Enjoy.

–TPM

* * *

With the announcement of the four-processor AS/400 model E95 and the unveiling of the long-awaited 9337 disk arrays, IBM has put the AS/400 line squarely into the mainframe arena. In doing this, IBM risks a wave of purchase deferrals by prospective ES/9000 customers, whose systems lack some of the AS/400’s key processor and disk file features. But that is the least of its problems. With its September announcements, IBM has inadvertently raised questions it is not yet prepared to address.

At the core of IBM’s dilemma is its announced and partially completed plan to more explicitly divide itself into autonomous units. While the reorganization provides IBM with a pretext for massive redundancies, it does not in any way address the cravings of the market nor assure a better financial future for shareholders. On the contrary. An IBM lined up behind the AS/400 would serve both constituencies far better.

I do not believe today’s AS/400 is a complete and truly general-purpose computer system, but I am confident it has within it–far more than does any other IBM product line–the seeds of such a computer family. The AS/400’s virtues are not only of great value to customers; they also would provide IBM and its shareholders with a steady and growing stream of profits . . . much the way the System/360 and its descendants did from the mid-1960s until the late 1980s.

What the AS/400 has is a range of completely compatible processors using a single proprietary operating system and data base management system. Because the AS/400 environment is completely under IBM’s control, it is consistent and capable of forcing consistence on software developers; applications developed by commercial programming companies and customer’s technical experts alike can share data. Now the AS/400 also has facilities that promise to make its reliability, availability and serviceability the equal or better of any commercial system . . . and at a lower cost when all the applications development and personnel training costs are properly counted.

But the AS/400 still fails to serve all the needs addressed by other IBM (and non-IBM) computers, and IBM could not make its best computer its only computer without addressing the product line’s shortcomings. To that end, I decided to come up with a list of the things IBM could provide to make the AS/400 a nearly universal computer system. The list is shockingly short:

A single user machine/workstation. The E01 would have eight to 16 megabytes of main memory and one to four gigabytes of disk capacity. It would include basic ports–printer, modem, mouse, keyboard, display and LAN. It would run the graphical or character versions of OS/400 (see below). And it would cost as little as $5,000 with a color display and basic systems software.
Graphical OS/400. Extensions to OS/400 that permit terminals with graphical user interfaces to function in complete harmony with a host AS/400 of any size. The software would be a compatible extension of current software so all existing terminals could be supported. . . at least when they run applications written for a character-mode terminal.
Virtual data base extensions. All the promises of peer-to-peer communications should be met and then extended to allow customers to build integrated groups of machines that are as large as any organization might want to install. An AS/400 linked to other machines should be able to ask its neighbors for information and instruct these same neighbors to accept information. Each machine in a cluster of AS/400s should be able to track the state of all data bases on all neighboring machines.
A vector processing and high-speed computing coprocessor for each model. AS/400s are poor at math and other computationally intensive tasks. There is no reason for customers with basic transaction processing applications to pay for this feature, so if IBM cannot come up with a way to add it to the machines at low cost, it could charge for the capability. I would rather see the processors beefed up to include a math coprocessor and thereby tempt all customers toward putting every type of application on an AS/400, but a separate and reasonable charge for the feature might be acceptable.

That’s it.

Coupled with some migration aids for other customers–ES/9000 users who can see the handwriting on the glass house wall; RS/6000 users who have realized just how unstandardized the Unix world really is; and those with networks of personal computers who still remember when computers worked nearly all the time–IBM could not only simplify its byzantine structure, but get another 20 or 25 years of renewed growth. After that, I suppose, it will be time to re-examine computing once again and replace–only if it is really necessary–whatever technological descendants the AS/400 has spawned.

* * *

Well, we got the vector processors with the Power6 chip. And IBM did deliver DB2 Multisystem in 1995, but never did much with it. There’s still time to blow the dust off of that and do something interesting. And we are still waiting for native graphics and an entry development machine, plus a whole bunch of other things I have learned to ask for.