There was something deliciously ironic about IBM's recent announcement that the S/390 will now be able to support numerous copies of Linux running under VM. Yes, Linux can also run native on the mainframe, or in an LPAR (logical partition), for those who like life to be really exciting. But I suspect the VM solution will have a particular attraction for users who see the S/390 as the place to centralise and consolidate their growing farms of distributed intranet and Web servers. After all, VM is a very mature 'hypervisor' (about 25 years worth of maturity), and would offer huge flexibility with multiple Linux images.
VM and Linux: what a fascinating pair. The one staunchly IBM-proprietary, reliable, safe, unfashionable, the domain of a declining but extremely loyal group of CMS developers; the other young, "free" and chic, the primary focus of Internet service providers and up-coming software specialists. As different as chalk and cheese, yet, in this case, each dependent on the other to provide the software environment that users now need. VM offers the tools, the scalability, the performance management; Linux brings new applications and potentially vast opportunities: ignore it at your peril!
Running Linux under VM is just one of the paradoxes facing today's enterprise systems community. The mainframe remains the most scalable and available platform around (even if competition from the alternatives is very fierce); but much of this advantage has been built up using products and tools that are now being relegated to back-offices and specialist maintenance teams. Out with COBOL, PL/I and the like; the new world belongs to Java and XML.
The juggling act that IBM has had to perform over the last few years is to retain the strengths of its highly resilient, established operating platforms and tools, while opening them up to new types of application and innovation. Not that Linux, or the new HFS support in OS/390 Unix System Services, or the various nascent methods of supporting NT applications on the mainframe, are key sources of revenue - well, not yet anyway. 2300 downloads of mainframe Linux (at the last count) from the Marist University site is not at all bad, but free software tends to attract more than its fair share of keen shelfware collectors as opposed to serious users.
Nevertheless, each one of these steps allows IBM not just to broaden the appeal of the S/390, but actually to play down the staid image of its top-end machines, and to increase the number of options on offer to its most influential customers.
IBM's suggestion that Linux will make the mainframe 'cool' is a little tongue in cheek. But it does help to achieve a number of important goals: further reinforcing the strength of the (largely anti-Microsoft) Open Source Software movement; raising the profile of the S/390 in universities, which are producing generations of computer science graduates with little or no mainframe awareness; and further standardising the range of software components that are available across IBM's four, culturally different hardware ranges.
For IBM, as indeed for the rest of the industry, Open Source represents a very different pricing model, and one which will ultimately have a profound knock-on effect on more established areas of enterprise software. The idea of a free mainframe operating system would have seemed incredible just a few years ago, but IBM is well positioned to make its money instead from tools, from integration with existing CICS, IMS, and DB2 sub-systems, and of course from support. IBM Global Services will provide the back-end support to distributors SuSE and TurboLinux, and this is no small consideration. Linux may be the most polished and widely debugged OS available, but consolidating hundreds of business-critical Web and intranet servers onto the mainframe will place unprecedented demands on the system.
And while the Open Source philosophy will gradually drive other software costs down, there is no shortage of service-based revenue to be had.
So if you thought that IBM's desire to win street credibility for the S/390 had finally caused it to lose its keen financial instincts, think again.
Mark Lillycrop is director of research for Xephon. firstname.lastname@example.org