All Things to All People

Can the AS/400 be all things to all people? More important is the question whether IBM should even try to make the AS/400 all things to all people. I certainly don’t know what IBM is going to do—especially since I am pretty sure IBM doesn’t know what they are to do either.

If we look at recent IBM announcements, they seem to be incapable of marketing the AS/400 as a premier platform for running core business applications – the AS/400’s traditional power base – instead concentrating on showing that the AS/400 can run AIX applications (with PASE) and the Linux OS (in an LPAR partition).

Picture an ultra-reliable server with the ability to run just about any application written by man.
I guess it is not necessarily a bad thing that the 400 can run AIX apps or Linux in a partition, but don’t you think the AS/400 community would be better served by a sound strategy for bringing new users to the 400; a strategy that didn’t involve telling them the non-400 things the platform can do.

The way I see it, an AS/400 that can do all these things is no more than a glorified Intel box. Unless, of course, you sell the traditional strengths of the AS/400 as the main reason to buy, followed by the fact that it has tremendous application execution flexibility.

There has been talk about merging the AS/400 and RS/6000 platforms—which makes a lot of sense since the hardware of the two systems is somewhere near 75% common. They use practically the same chip (the AS/400’s PowerPC AS has a few extra instructions) and are produced on the same production line in Rochester. Only at the end of the manufacturing process do they diverge to become AS/400s or RS/6000s.

Pardon me for being an AS/400 cheerleader, but I think that a merger of the two platforms would be a mistake. Instead, I believe it would make more sense if the AS/400 subsumed the commercial uses of the RS/6000, and the RS/6000 lived on as a scientific computing platform. Certainly you would continue to sell the RS/6000 workstations—in addition to the RS/6000 SP supercomputers.

If you believe IBM, AIX applications recompiled to run on AS/400 using PASE suffer only minimal, if any, performance penalty. At first I wasn’t convinced of PASE’s ability to execute AIX applications on AS/400 with little or no performance hit, but after reviewing the architecture of PASE, I am willing to give IBM the benefit of the doubt. Further cementing my belief that PASE can deliver performance as advertised is a conversation I had with a representative of SPSS, one of the first ISVs to “port” their application to the AS/400 using PASE. When asked point blank about the performance of their application under PASE, they made it clear that there was no appreciable performance penalty running SPSS on the AS/400.

The only other major hurdle to running any and all AIX applications on the AS/400 is library support, which if not complete at this time, could certainly be made complete in the near future.

Now comes the question of Linux. If successful – and that is a big ‘if’ if you ask me – Linux will spawn a flood of applications, just like the introduction of Windows NT did in the early nineties. For this reason, bringing Linux support to the AS/400 cannot hurt the platform as long as people are not detracted from developing native AS/400 applications.

Originally I thought that IBM would bring Linux application support to the AS/400 through PASE, not full Linux OS support through LPAR. In either case, the ability to run Linux applications on the AS/400 will only serve to strengthen the AS/400’s position as a server.

In a really far-out scenario, IBM could even partner with Microsoft to resurrect the Power PC version of Windows NT, and bring that to the AS/400 in a LPAR partition. Certainly this is a fantastic scenario, but it is technically possible and desirable for many reasons.

With the ability to run native AS/400 applications, AIX applications, Linux and Windows NT, the AS/400 would be the ultimate server consolidation platform. Picture an ultra-reliable server, with power redundancy, shared disks, centralized backup capabilities and the ability to run just about any application written by man. This is a very pretty picture indeed.

The operation of all these different applications and operating systems would be monitored and managed from within OS/400, simplifying systems management. You could freely assign disk space as needed among the different operating environments and you could partition the system to allot as much power as needed to any specific application or OS.

Science Fiction? Maybe. Technically possible? Absolutely. The only question is whether IBM has the vision to see what needs to be done, and the kahones to pull it off. Hmmm… maybe the AS/400 can and should be all things to all people after all.

Related Editorial:

  • IBM's Unix Migrates to Monterey!
  • AIX Applications—PASEmaker for AS/400?
  • Things Change

    Related Information:

  • IBM Linux Page (new window)
  • IBM AIX Page (new window)