IBM Launches Autonomic Computing Group

What IBM calls "autonomic" computing is a response to the complexity that saps the potential productivity gains from the latest technology, said Alan Ganek, a 20-year IBM veteran who is vice president of the new autonomic computing organization.

"Deployments are incredibly complex and almost acting at odds against the improvements we've had in price-performance," Ganek said. "Chips are getting cheaper and cheaper, but the way we put them together is more and more complex. That is an obstacle."

In the short term, IBM is working to make software products like DB2 and Tivoli more self-tuning even as the vendor's hardware groups tackle similar problems.

How much sway Ganek will have over IBM's disparate product groups is unclear, however. Frank Gillett, principal analyst at Forrester Research, likened the autonomic computing initiative to IBM's Pervasive Computing effort, which also extends across different fiefdoms at the company. The head of that effort has a mandate but is not in charge of a lot of products, per se, Gillett said.

IBM's Ganek also said it's critical that this not be an IBM-only effort.

He said the company is pursuing an open-standards road map, and that IBM and Microsoft have already blessed the Open Grid Services Architecture game plan and are proponents of other Web-based standards that will come to play in autonomic computing.

"The general idea is that any solution limited to a specific hardware technology architecture, whether it's mainframe or Intel, doesn't solve any customer problems," said IDC analyst John Humphreys.

IBM's Ganek said the company wants to work with many other vendors to broaden the autonomic computing effort.

"We're building a reference architecture and approaches that will be available to other companies. Lots of other people's software drive our servers," Ganek told CRN.

What will be key is raising the lowest common denominator so users get the most out of their IT spending and integrators can spend more time and resources extending capabilities rather than getting existing systems to work right, he said.

Everyone agrees the complexity problem is real. "A lot of [computing capability is wasted now. I have to buy double, triple maybe even five times the stuff I need because I have to buy for peak volume, not average volume," Ganek noted.