IBM Corp. is investing billions in research toward developing self-managing, self-configuring, and self-healing servers that require little human interaction.
The company is devoting 25 percent of its server research and development budget to technology that will free IT professionals from mundane and often difficult, time-consuming tasks whose outcomes are hard to predict, such as capacity planning and server failovers.
While Compaq Computer Corp., Hewlett-Packard Co. and Dell Computer Corp. each have self-management server technology, none of the companies would comment on the amount they invest in such technology.
Called Project Eliza, IBM's "intelligent" hardware and software will in the next few years appear across IBM's server family - from mainframes, AS/400s and RS/6000s to its Intel-based boxes. As soon as next year IBM servers may be able to:
- Activate built-in redundant components when failures occur.
- Automatically balance bandwidth or application capacity when necessary.
- Monitor for intrusions.
- Cluster with other servers on the fly to balance workload and for failover, redundancy and increased availability.
- Automatically configure and install operating systems, applications and data.
The amount of data on the Internet is going to continue to grow at an explosive rate, says Mike Nelson, director for Internet technology and strategy at IBM. "There's no way we'll be able to deploy enough servers and make use of that data if we don't make the server easier to install and manage. You can't train and hire enough IT professionals to take care of that data if we're using the systems we are today."
Users have already glimpsed a few of the redundancy features IBM and other vendors have built into their servers for failure. IBM's xSeries (for-merly NetFinity) servers are equipped with a technology dubbed software rejuvenation that views the processes running on a server and restarts Windows NT and 2000 servers before they fail.
IBM xSeries servers and Dell PowerEdge 6400 and 6450 boxes are already outfitted with Chipkill technology, in which data is automatically redirected from failed memory chips to segments of memory that are not faulty. Perhaps the most recognized self-management features are those Compaq ships on its Proliant servers.
"Our servers can predict many failures and report them using a variety of means," says Terry Roedecker, senior network administrator at MidFirst Bank in Oklahoma City, Okla. "Although [self-managing technology] is still in its infancy, it has saved us more than once. We've had drives report as being ready to fail. By knowing that ahead of time and getting them replaced, we've averted possible downtime on the server."
An eye on capacity
Another IBM Eliza project is capacity-balancing technology that could be used by a company for its Web sites. For example, if servers in one site had too much traffic to handle, traffic could switch automatically to servers in another site. Or the servers in one site could be clustered, aware when another server in the cluster was overloaded or failing and take over the capacity of the failed server.
"Capacity is a problem we face every day," says Roedecker. "We have servers with more [capacity] than they'll ever need and others that never seem to have enough. Having the ability to allow the servers to automatically remove [capacity or bandwidth] from one device and add it to another would be a [boon]."
While capacity management will be a time-saver, Rocco Esposito, CTO of window-covering manufacturer Hunter Douglas in Upper Saddle River, N.Y., says his staff spends about four hours per day monitoring servers, 30 hours per month monitoring capacity and another 30 hours per month installing and configuring server software. With a self-managed server such as that from IBM, Esposito could eliminate most of this time and free his staff for more important tasks.
"This is an activity that is spead across the company," Nelson says. "There are pieces that are going to be in the research, software, and servers group."