The servers based on IBM’s Power6 processor architecture are going to be delayed. A launch was scheduled for mid-2007, but now even this vague timetable is apparently too optimistic. But, actually, who cares?

In the world of enterprise IT, the importance of processor architecture – and hardware in general – is in decline. IBM still seems keen to use its intellectual property rights (IPR) as a market bludgeon, but would do well to remember that undue reliance on hardware has caused it problems in the past.

IBM’s previous high-profile addiction was to the PC architecture, which it tried and failed to keep to itself. The subsequent growth of the PC clone market caused a complete breakdown in its ability to compete with younger, more nimble rivals.

Now IBM is trying to defend its ownership of mainframe hardware architecture – and derivative hardware sales – by controlling ownership of the z/OS operating system. Its tactics are summed up in an on-going legal spat with PSI, a code translation specialist that worked with IBM to port z/OS to run on the Itanium processor.

When IBM dropped Itanium, PSI offered its technology to all comers, with HP showing significant interest. So now IBM is using legal measures to try to stop PSI from creating a situation where z/OS users might move kit and caboodle to an Itanium box.

But giving users a path to a cheaper hardware alternative is hardly the end of the world, not least because it may open up new demand for IBM software. That IBM is using technology from Transitive to port Linux and x86-based applications to future pSeries boxes suggests that the company is well aware of the potential.

Any enterprise with an existing pSeries infrastructure may want to run an x86 application on a pSeries machine, but there is an equal chance they may go the opposite way. Lots of firms now perceive the mission-critical reliability of z/OS back-office applications as a great theory, but an expensive reality. It might actually make better financial sense for them to run the OS on Itanium or, Lord preserve us, x86 boxes.

For vendors, the real money is not in the hardware architecture, but in providing the knowledge base required to keep the IT infrastructures supporting large, complex business systems running.

So squeezing the IPRs to a specific processor architecture to support the necessary software (unless there is some extremely perverse technical reason why it must run on that architecture alone) is actually a rather pointless and possibly dangerous exercise. And one that could cause serious health problems for IBM and its end-users alike.