he devilish complexity of computers and — worse — networks of computers, has long been the bane of the engineers who try to keep corporate computer systems up and running. And the complexity has exploded in recent years with the rise of the Internet.

Computer scientists are tackling the challenge with new tools and fresh thinking. One approach is to borrow ideas from biology. The human body, after all, can be seen as a phenomenal processor of information complexity. The chemical units of DNA — A, G, C and T — are not so different, really, from the digital code of 1's and 0's.

Advertisement

I.B.M. is taking a page from biology with a planned announcement today that it is creating an "autonomic computing" unit. The group will be headed by Alan Ganek, former vice president for strategy in the research division. The autonomic computing effort will span the company's hardware, software, services and research operations, and analysts estimate that I.B.M. will be spending $500 million a year on the initiative.

The group's goal is to develop technology that will do for corporate data centers what the autonomic nervous system does for humans — automatically handle basic functions, like breathing and digestion, or in computing's case, networks that diagnose bugs and fix them automatically and data centers that automatically figure out the best way to handle changing workloads.

Better hardware designs are part of the answer to overcoming computing complexity. But clever software of various kinds is probably the most important technology.

Software that can log a network's behavior and learn from it is one promising area, Mr. Ganek said, helping to make systems that are predictive and adaptive. Another important tool will be software that enables people to state business objectives with graphic or text representations — not in a complex programming language — and have the network figure out how to tackle the problem.

Mr. Ganek emphasized that pursuing autonomic computing will be a step-by-step process that will take years. "There is not some new product that will suddenly make things autonomic," he said.

Other companies are also addressing the challenge of complexity in corporate data centers. Sun Microsystems' N1 initiative seeks the same goal as I.B.M.'s autonomic effort, as does Hewlett-Packard's Utility Data Center project. "All the big vendors are going after the complexity problem," said Amy Wohl, an industry consultant. "But I.B.M. has been working on this longer and has more researchers focused on this than anyone else."  STEVE LOHR