There's a new skirmish between the old and new schools of IT architecture. The old school calls for matching the system used to the workload at hand -- we've been doing that for decades. The new school -- the cloud school -- argues for massive replication of commodity hardware, which can be sliced up, shaped, and reshaped to handle any workload thrown at it.
If you thought there was going to be a smooth and rapid transition from the old school to the new one, recent evidence suggests neither vendors nor IT pros see that happening. Now there's some vendor gamesmanship going on, and reading that right will tell you a lot about the vendors you're working with and whether the transition will happen on your timeline or theirs.
As for the gamesmanship, Oracle said last week that it'll stop developing its database and other software for Itanium because Hewlett-Packard and Intel aren't committed to the future of systems based on that chip. Oracle, it appears, still believes in selling only workload-specific systems -- mostly its own. Both HP and Intel, meanwhile, deny any intent to dump Itanium, and if you consider the size and complexion of that business -- particularly on HP's part -- Itanium will have a long horizon. But we didn't need the yelping from Oracle to see that workload-specific systems are alive and well. IBM's entire product line is a nod to that thinking. Big Blue continues to develop the z (mainframe), p (proprietary Unix) and x (x86-64) series of servers, and -- at least until Oracle declares otherwise -- there's no evidence IBM is giving up on any of them.
For its part, HP knows a thing or two about what it takes to sunset a processor architecture. It has killed more chip lines than most any other company has started. Some of the most notable ones include Alpha, which it got when it acquired Compaq (and Compaq got from Digital Equipment) and its own PA-RISC family. The upshot is that killing processor lines is an expensive and unpleasant prospect, and it seems unlikely that HP would do so with Itanium considering it has no alternative for its own proprietary Unix, HP-UX, a still lucrative though shrinking segment of the high-end server market.
IT transitions of any large scope -- whether they're away from an IT architecture, like workload-specific computing, or away from a processor family, like Itanium -- take a decade or more. So vendors that have the interests of their customers at heart let the transition happen at its natural pace. Vendors with limited offerings will try to force the issue - as Oracle is doing here. Its actions are dictated by the strange hand it has dealt itself, in that it wants to own the last generation of IT architecture. Sun hardware running Oracle software is all about workload-specific design, and while that's a high-margin market now, it's one that will slowly but surely give way to cloud computing.
Oracle is trying its best to force HP into supplying only commodity-level cloud products to its customers well before HP would have had any intention of heading down that road. The big loser here is likely to be customers that still want Oracle software running on HP hardware. While it's harder to replace a strategic software vendor than a key hardware vendor, customers should opt to keep the vendor that's keeping their interests in mind. If it were my IT shop, I'd be looking into an exit strategy from Oracle.
Art Wittmann is director of InformationWeek Analytics, a portfolio of decision-support tools and analyst reports. You can write to him at [email protected].
To find out more about Art Wittmann, please visit his page.More than 100 major reports will be released this year. Sign up or upgrade your InformationWeek Analytics membership.