Technology News Technology News

Another Perspective


Crichton was merely a servant of the Lasenby family in Victorian England.  But when the Lasenbys were marooned on a desert island, it was Crichton whose gifts came to the fore.  Before long, Crichton was in charge, respected by the men and admired by the women.  Then the Lasenbys were rescued.  They soon reverted to their former places, geographically and socially.  Will this be the story of Linux in IBM's Proprietary Empire?  IBM seems to think there's an alternate ending to the tale.

It's beginning to look like Linux is slated to be the primary operating system for new applications, particularly those with Internet functionality, on its two proprietary server families, the zSeries and the iSeries.  Basically, IBM is telling customers to do everything that uses a GUI to use Linux as a host environment.

So, while Linux may have begun its life in the zSeries and iSeries worlds as a servant, serving Web pages and performing other Internet chores, like Crichton in the Lasenby household, it could end up in a position of dominance.  Unlike Crichton, Linux on IBM's proprietary systems may never have to relinquish its powerful role.

Just a few years ago, it would have been impossible for anyone — possibly even anyone in IBM — to guess how important Linux might become to mainframe and iSeries shops.

The point of view in James M.  Barrie's social comedy The Admirable Crichton is that people who have all the requisite skills to lead in a state of nature might not be able to exercise their gifts in the decidedly unnatural environment offered by civilization, in particular Victorian England.  And IBM's point of view is that its systems must adapt to the nature of a marketplace that has embraced Windows, the Internet and GUIs, spurning the green screens and strictly character interfaces of the old EBCDIC Empire.  That there are quite a few citizens of the old IBM world who are pretty happy, just as they are, is not sufficiently important to change Big Blue's corporate mind.

James M Barrie
James M Barrie
Made Peter Pan buy bed pans

Barrie's observation about his culture and times went over pretty well, it turns out.  The Victorians were secure in their empire's strength and enjoyed entertainments that raised questions about their realm.  IBM doesn't feel that way about its legacy systems.

Barrie's approximate contemporaries (who were a bit older) Gilbert and Sullivan certainly enjoyed great success, with light operas chock full of social humor.

Barrie became as famous as Gilbert and Sullivan, and, very possibly, just as rich, mainly because of the immense appeal of his Peter Pan.  In fact, he made so much money from Peter Pan that in the end he really didn't need any more.  Without children of his own, he assigned his rights in the property to all the children of England: In 1929, eight years before his death, Barrie gave Peter Pan to the Great Ormond Street Hospital, which has an outstanding record in pediatric medicine.  Through a combination of legal trickery (in a good cause) by former British Prime Minister James Callaghan and changes in the copyright laws of the European Union, the copyright on Peter Pan will hold until 2007, giving the hospital another five years to benefit from Barrie's largesse.  We would be happy to say just how much Great Ormond Street collected in royalties, but one condition of Barrie's gift to the hospital was that it must keep its income from Peter Pan secret.

Peter Pan will almost certainly get a boost from new productions to commemorate its hundredth anniversary in 2004.  The Admirable Crichton, however, is not likely to make a comeback.  Or maybe it already has.  There are two films entitled Swept Away based on the play.  The 1975 version is pretty well regarded by critics, but the current release, starring Madonna, has not proved to be a crowd pleaser at all.  Maybe it's something other than the acting and the direction of the latest version; although these are the aspects of the film that have caught the brickbats.  Maybe the idea behind the story is out of step with the times in the world at large, just as it is at IBM.

IBM seems to think that just about all legacy software — and particularly legacy software that provides islands of computing that talk to green screens — might have to be left on islands in the corners of customers' installations.

Rather than extend and, where necessary, rework its proprietary operating systems and middleware to include the full range of Unix and Linux services and features in a native fashion, IBM has given all of its server families the ability to support Linux.  (OS/400 supports AIX runtimes in the OS/400 PASE environment, but there are hitches to its use.) Linux is a natural host for Internet-type services; in fact it is the most popular Web hosting environment (by number of domains, not the only measure).  The related Apache Web server (which also runs in many other environments) is, according to Netcraft, by far the most popular software of its kind.

Supporting Linux requires no work whatsoever in the case of Intel-based xSeries servers, but IBM is doing a lot more than the basics.  It is bringing advanced technology, such as clustering, to the Linux world, hoping, quite reasonably, that it will get a very healthy chunk of the Intel-based server business that its contribution generates, even if some of its gifts end up helping competitors to sell Linux systems.

Linux is more a matter of policy than technology in the pSeries line.  Porting Linux to platforms that are already big winners in the Unix market is not a big challenge for IBM.  Adjusting to a market that might prefer Linux to AIX on pSeries might prove to be a bit more difficult.  But IBM could well discover that Linux on pSeries is just what the market wants, particularly if Intel's 64-bit chips turn out, as the wags have it, to be the Itanic.

In mainframes, IBM had to do a bunch of work to make Linux a practical option.  First, it had to get Linux to run on the boxes, or run in a limited fashion.  (There is still no Linux support for tape and various other devices.) Linux on a mainframe is sold on dedicated engines that have special microcode but cost less than the same engines licensed for the proprietary OS/390.  And VM, which enables a zSeries machine to run multiple Linux system images and which couples Linux to legacy processes and devices, had to be reworked and repriced to make the whole proposition reasonably attractive.

And now there's the iSeries, on which blossoming Linux support is starting to look a lot like the offering that IBM has made in the mainframe world.  In some ways, Linux on the iSeries will be easier to understand and use than it is on the mainframe.  For instance, OS/400 will play two parts in this drama.  IBM does not support multiple proprietary operating systems in the iSeries the way that it does on the zSeries.  As a result, the same environment (or various versions of the same environment) on an iSeries box can do what IBM uses both VM and OS/390 to do on a zSeries.

Whether the iSeries variation of the strategy turns out to be more effective in the market than the zSeries plan will depend on the reactions of an audience that is as different from the mainframe base as the Swept Away audience in 1975 is from today's moviegoers.  What plays well to one crowd might play better or worse to another.  iSeries users might not have sufficient resources to become adept at using Linux; they may need all the talent they have just to keep their OS/400 environments running productively.

Picture panned

The essence of the IBM Linux plan for proprietary systems, as it has unfolded so far, is that users should lock up their applications and leave them in their established environments.  But when it is time to bring data, and perhaps some functionality, to the Web or to intranets, the way to go is Linux.  In addition, any work that these users might be doing (or wish to do) on Unix, Linux, or Windows boxes ought to be done using Linux on their central zSeries or iSeries machines.

This is a lot more than a notion.  IBM is so serious about this concept in the mainframe world that its newest mainframe, the z800-0E1, can only be acquired in one basic configuration.  This configuration has two active engines.  One provides 40 MIPS of traditional mainframe power, half the power of the prior low-end machine in the z800 line.  The other engine is strictly for Linux and VM (restricted to managing multiple instances of Linux); it cannot be used for ordinary production jobs by VSE operating system users, who constitute nearly all of the low-end mainframe base.  The Linux engine is a full one, and if the microcode let it run OS/390 (which it does not), it would run at 192 MIPS, or nearly five times the speed of the Linux engine.  That says an awful lot about IBM's vision of the future for VSE users.  It's one in which the legacy software is left on an island in the midst of an ocean of Linux computing power.

The plan for the iSeries might soon resemble this situation very closely, perhaps to the chagrin of OS/400 shops that have bet their existence on a future that looks very much like the present.

In the case of small mainframe shops, IBM has made it clear that it will no longer offer small mainframes that do not include big Linux capability.  Customers who want only small mainframes can only buy used machines (or little mainframes emulated under SCO Unix on Intel platforms).

How would low-end OS/400 users react to the same sort of move in their corner of the world?  They might not love it.  But, as is the case for their counterparts in mainframe country, there sure isn't anything they can do about it.  There is only one source of platforms that support VSE, VM, and OS/390.  There is only one source of platforms that support OS/400.

Smaller shops with pure legacy workloads might just ignore what is going on around them and stick to their islands of traditional computing.  This has happened in a dramatic way on at least one other island, this time in the (slightly) North Pacific; Crichton was in the South Pacific when he had his day in the sun.

Sgt.  Shoichi Yokoi, a Japanese soldier, held out on Guam until January 24, 1972, by living in caves and generally not raising a ruckus.  When he was finally discovered, and was persuaded that Word War II had ended, in 1945, he decided to go back home to Japan.  At first he was treated as a cultural hero, and deservedly so.  He even got married (to a wife he found by advertising) and settled down.  He received many gifts, so his life was pretty easy.  But he never quite adjusted to the modern Japan, and when he got into the media griping about the loss of traditional values (such as obedient women), that was the end of his cash flow.  He had, however, amassed enough to live on, and stuck around until 1997, reaching the age of 82.

A legacy kind of guy

A lot of shops with small mainframes and, very possibly, tens of thousands of OS/400 sites might yet be putting up Yokoi's picture in their server rooms.  He was their kind of guy.

But their reaction is not the only possibility.  Some users might keep their humming legacy systems for as long as they can and then disappear, like Judge Joseph Force Crater.  On August 6, 1930, on the way to a play called Dancing Partner, which was running in New York, Judge Crater just disappeared without a trace.  Some say bad people he had dealings with put him to sleep with the fishes.  Others say he had another life set up somewhere, maybe a nice island, and that this was his time to go there for good (or whatever he went there for).  The last fact established by the authorities ends with Judge Crater cashing a very large check.

That's the kind of ending IBM is shooting for with its Linux plan: happy, and if nobody quite understands just what happened, so what?

— Hesh Wiener October 2002

Copyright © 1990-2019 Technology News of America Co Inc.  All rights reserved.