Rumor Du Jour: i5/OS on Other Platforms? Not!

July 23, 2007 Timothy Prickett Morgan

The midrange press was abuzz a little last week about the prospect of IBM porting the i5/OS operating system to other platforms. While this would be an interesting development, and one that would quite possibly breathe some new life into the i5/OS and OS/400 platform, it turns out, alas, that such talk is not true and is the result of a misunderstanding of some comments made by an IBM executive at a user group meeting.

It all started when Elaine Lennox, the vice president of marketing for what has been called the System i division until last Thursday and which is now called the Power Systems division (or at least part of it is called that), was speaking at the OCEAN Technical Conference in Irvine, California, on July 16. OCEAN is one of the largest regional user groups, serving Southern California, and it usually draws a pretty big crowd. Somehow, Lennox’s comments got misinterpreted by people at the conference, which were then passed on to members of the trade press. Before you know it, people were thinking IBM was going to port i5/OS to other platforms.

According to Lennox, this is what she said that set everybody off: “Looking forward it was quite possible that we’d offer i5/OS on other platforms–such as, for example, why not put i5/OS on blades into BladeCenter one day in the future?”

Instead of hearing that as putting a Power6 chip into a blade form factor and allowing i5/OS to run on it, as we have reported that IBM was planning to do for several years now, somehow some people listening to Lennox speak heard “other platforms” instead of what she actually said. So no, IBM was not talking about the possibility of porting i5/OS to other hardware platforms, but really making i5/OS available on a future Power6 platform in a form factor that should have long-since been able to run i5/OS.

(And to be precise, as we report elsewhere in this issue, Mark Shearer, formerly the general manager of the System i division and now the senior vice president and business line executive for IBM’s new Power Systems division, said in an interview last week with me that this Power6 blade server supporting i5/OS would be available in the fourth quarter.)

If any customer base in the world understands the inherent value proposition of an integrated platform, it is the i5/OS and OS/400 customer base. Having integrated the operating system, database, middleware, and development tools into a single stack and pulled storage under the covers, too, the next logical thing would be to internalize the entire network, including switches and routers and other gear. If anything, the OS/400 platform from six years ago should have been the first blade server form factor, much as AS/400 machines were the first rack-mounted servers. Yes, it steams me that the i5/OS and OS/400 platform used to lead and now it follows. But, in this case, the special memory tags and some of the instructions for running OS/400 and i5/OS are not on the PowerPC 970 processors used in IBM’s BladeCenter machines, and the Power4 and Power5 family of 64-bit processors, which do have these memory features and instructions, were far too hot to put on a blade server. By moving to 65 nanometer processor manufacturing technology and by, presumably, gearing down the clock speed and making use of other power-down features that shut off unused portions of the Power6 chip, IBM can get an i5/OS-compatible processor into a blade form factor.

While I am thinking about porting i5/OS to other platforms, it is worth bearing in mind that there is absolutely nothing that prevents IBM from doing such ports. When IBM made the transition from CISC to RISC chips in 1995 and into 1996, the mix of PL/1, assembly, and heaven only knows what other kind of code was ported to a mix of C++ and a smattering of Java. IBM could port OS/400 or i5/OS to any platform it chose to, but it might have to do a lot of optimizations to make certain features (like virtualization) work properly. It might have to force a recompile because of memory and other instruction issues, too, or have another emulation layer between the operating system’s microcode and the iron to make i5/OS think it is running on Power when it really isn’t. The fact is, IBM chooses to keep i5/OS on its flagship, commercial Power platforms. It is not a foregone conclusion, and that is what is frustrating to those of us who want IBM to run this platform on less expensive iron than it currently does. If the new Business Systems division, which inherited the System i 520 and 550 as well as the 515 and 525 servers doesn’t understand that, the i5/OS and OS/400 community is happy to explain it to its new leaders. You can rest assured I will do my part in the dialog.

There is, of course, a new alternative, and one that IBM was probably smart enough to invent itself and maybe it did in its research labs but never had the insight or nerve to deploy. Total emulation, and with the amount of processing capacity that is available in chips today, this is a real option. Just ask Transitive, the makers of the QuickTransit emulation environment that lets PowerPC-based applications run on an X64-based Mac from Apple Computer. IBM is well aware of QuickTransit, since it can be used to port mainframe applications to Linux or Unix boxes as well as porting one kind of Unix or Linux to a Linux or Unix on another hardware architecture without forcing customers to recompile those applications. IBM is using QuickTransit to let Linux applications compiled for 32-bit X86 processors run on Power chips running Linux in an emulation environment that it is giving away to customers called the System p Application Virtual Environment, or PAVE.

IBM has lots of options to get i5/OS applications onto other chip architectures, if it wants to do that. This much is clear. And so far, IBM has not been inclined to move in that direction.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search