Pushing 'smart' computing

ATTEMPTING TO lead customers into the promised land of its on-demand computing environment, IBM is placing a huge bet that a weaving of autonomic, grid, and utility computing will convince them to make the journey.

Free IT resource Virtualization Insights from Top Experts - Learn how virtualization gets real! Sponsored by Dell Free IT resource Try Sun servers, workstations and storage products free for 60-days. Sponsored by Sun Microsystems

IBM believes the right combination of these technologies, to be steadily embedded into all its core products, ranging from chips to server software, will take the pain and expense out of integrating and protecting disparate islands of corporate data through the creation of the next generation of intelligent and adaptable infrastructure.

Although on-demand computing is still several years away, IBM is putting several pieces in place now while aggressively pursuing its autonomic agenda, delivering versions of Tivoli, DB2, and later this month, WebSphere. Each comes with a fistful of self-managing, self-healing, and self-optimizing capabilities. IBM also will deliver new autonomic abilities in the next major release of its pSeries and iSeries servers due in 2004 that will allow them to automatically access and manage virtual resources, much like its zSeries mainframes.

IBM will evolve the autonomic capabilities of middleware, including WebSphere and DB2, by infusing them with open-standard Grid protocols, according to Irving Wladawsky-Berger, vice president of technology and strategy for IBM's server group in Armonk, N.Y.

"The more difficult part of this is to make not just individual products autonomic but the infrastructure itself more autonomic. Then you can start managing distributed applications in a large enterprise in an autonomic manner," Wladawsky-Berger said. "By integrating Grid protocols into future versions of WebSphere, users can automatically carry out distributed resource management to make decisions about optimizing performance. They can determine the size of a distributed cache or how many copies of an application should be running to increase availability."

Despite what IBM officials claim is an obvious high return on investment, users and analysts would like to see more implementations of the technology before they become true believers.

"It is not clear yet how on-demand computing really works. This idea of computing as a spigot you turn on and off is new. I don't think OEMs, for instance, have figured out a way to economically wheel computing in and out of an environment yet," said Roger Kay, analyst at IDC in Framingham, Mass.

Analysts caution that IBM needs to deliver short-term benefits that emphasize autonomic solutions -- above and beyond the utility aspects -- with healthy returns on investment.

"IBM recognizes the importance of delivering products over the near-term that show benefits other than those associated with utility computing, particularly around cost reductions, through better utilization of their datacenters and reducing the head count of those eye-balling their systems," said Mike Gilpin, analyst with Giga in Cambridge, Mass.

IBM officials admit that, in these tough economic times, convincing people of the practicality and cost savings of autonomic computing will take time, but they are unflinching in their belief that it will become pervasive throughout most corporations.

"It will be a little bit like trying to convince people to wear seat belts, then to install airbags, and then to install multiple air bags. People will become convinced that safety and reliability are very important," Wladawsky-Berger said.

Most of IBM's successes in autonomic computing have been with universities and research organizations, but company officials said they are gaining commercial users, such as John Deere and DaimlerChrysler, with pilot projects.

One institution interested in IBM's autonomic vision is the Wharton School at the University of Pennsylvania.

After multiple failures involving a range of mixed servers, including one failure that involved a critical data service Wharton makes available to 73 other universities, school officials said they are ready to look to a more unified environment that has self-managing, self-correcting abilities.

"[Computing environments] have become so complicated. I want to take a look at some simpler solutions. It is interesting that IBM has arrived at this time with these solutions that would appear to make things simpler as well as reducing costs," said Gerry McCartney, CIO at Wharton, adding that the cost involved in buying autonomic products and services is not the most important factor in such a decision. "For most people in my job, money is secondary to providing a stable service. I could come back to my users and say, 'I saved you $200,000, but your e-mail system is a bit dodgy now.' I can't envision any place where they would think that was a good deal," McCartney said.

Some users are holding back on making a commitment until what they see as kinks are worked out.

But these users see some short-term benefits.

"I don't think many of these [autonomic] solutions are seamless. But what will sell this to most people is not that it works perfectly all the time but that it alerts administrators so they can respond to small problems before they get bigger," said Chuck Davis, vice president of engineering at a large producer of agricultural products in St. Louis, Mo.