RealTime IT News

IBM Aims for Self-Maintaining Servers

Forty some odd years ago, the U.S. brought together its best scientists with a goal straight out of science fiction. They intended to put a man on the moon. IBM Corp. (NYSE: IBM) Friday set out on its own moonshot: the creation of self-managing servers.

The initiative is Project eLiza, and Big Blue said it will devote billions of dollars -- 25 percent of its research and development budget for servers -- to the project, as well as five of its highest profile labs -- including Haifa, Israel and Boeblingen, Germany -- and hundreds of its top scientists.

"The ultimate goal of eLiza is to eliminate most, if not all human interaction with business computers and make global computing networks as easy to manage as today's kitchen appliances," IBM said.

"Our vision of the future and interconnectivity is a billion people interacting with a million e-businesses via a trillion interconnected intelligent devices," said Dr. Tom Bradicich, chief technology officer of the IBM eServer Group.

There's only one hitch in that vision. According to the U.S. Department of Commerce, in five years there will be a shortage of at least one million IT administrators. The future may hold hundreds of millions of people connected via wireless and other devices to the Web, driving trillions of transactions, but who will be maintaining the systems? Who will fix the servers when they inevitably go down?

Enter Big Blue's Project eLiza.

"We are responding to that and preparing for that with Project eLiza," Bradicich said. "We're moving into an era when computers are going to need less and less, and in some cases no, support from a human caregiver."

Under Project eLiza, Big Blue plans to unveil servers -- from the high volume Intel-based xSeries all the way across the board to the top-of-the-line zSeries -- with the ability to "heal" themselves by activating built-in redundant systems when failures occur, the ability to protect themselves with super-vigilant security technology, and the ability to configure themselves by installing operating systems and data automatically. As a step in this direction, Bradicich said its Intel-based servers will be hot-swappable by the end of the year.

"(In the future) even if you have the money to pay the IT professionals, they may not exist," Bradicich said. "Having computer systems that have the self-healing, self-protection and self-provisioning capabilities will be essential in many cases."

Some Project eLiza technology is already in place, including the Chipkill memory technology which can recover a chip memory failure, and Intelligent Resource Director, a server management environment that automatically allocates system resources for multiple jobs, according to demand. Software Rejuvenator, a part of IBM's current management suite which can predict system lock ups and take evasive action.

More of the technologies are slated for release this year, including eServer clustering -- which updates software and manages workloads across massive clusters with a single operation -- and self-healing technologies, like Chipkill, intended to enable server components to function for decades without failing.

Further down the pipeline are subsets of eLiza like Project Oceano, a working prototype of a server farm that manages itself. It brings machines on line or off line to meet changing demand. It installs operating systems and data, and it diagnoses problems and takes corrective measures. IBM said it does all these things -- without human intervention.

And in several years, IBM plans to unveil Blue Gene, the world's first supercomputer with the ability to heal itself.

While the company has undertaken this project with the coming IT administrator shortage in mind, it also recognizes that even if Project eLiza is successful in its goals, there will be times when an IT worker will have to service the machines. Therefore, according to Bradicich, the company is also working on remote management capabilities that will allow workers to manage the system from a computer or even a PDA.

"Not only are we trying to do it automatically, but the things that cannot be done automatically, we're removing the time and place dependencies," Bradicich said.

"I think its a vision," said Jonathan Eunice, an analyst with the Illuminata Group. "You can't expect that you're going to get the full shebang in one year or three years or even five years. It's a long-term project. Once you understand that, I think it's pretty exciting."

Not only is it exciting, but it may very well set a new course, a renewed focus on quality and reliability, for the entire server industry, according to Eunice.

"I think you're going to see widespread imitation of it or widespread reaction to it," he said. "(Others in the space) are going to have to step up too as a result of IBM kind of lighting a fire under the issue.

"It's not going to be unique. The vision of having no management overhead, everyone's got that vision. If a company like IBM starts down this path, you can be sure that other companies will, one, seek to license it, or two, imitate it."

But Eunice also pointed out that while many will seek to do the same things, IBM will have a significant advantage due to the size of the investment in the project and the intellectual property it will bring to bear.

"IBM's back as a top-flight competitor," he said. "It's totally changed the market dynamics. It's not good news for Compaq, it's not good news for HP. It's really become an IBM versus Sun (Microsystems) show."