Smart server management with eLiza

Bill O' Brien: IBM may be an institution, but it's been developing some stuff lately that could put it perilously close to messing with a cultural icon - Star Trek.

IBM's continuing development of its Project eLiza initiative to create self-managing systems could make it a star-date to remember. ELiza's predicative nature promises to turn manually run diagnostics into a thing of the past. That means curtains for a Star Trek mainstay: Captain Kirk asking Scotty for a Level 5 diagnostic check of the Enterprise's computer system. Let Kirk hook up with IBM's eLiza, and Scotty will be looking for a gig in another galaxy.

Vintage television humour aside, eLiza has all of the earmarks of being an earth-shattering piece of software. Actually, eLiza isn't a single software package -- it's a series of products that will create an autonomic system on top of your hardware.

The Electronic Service Agent, for example, is an enhanced version of the self-healing capabilities defined by Project eLiza that can remotely detect and repair problems with any eServer, in many cases without human intervention. IBM describes another part of eLiza, the Enterprise Workload Manager (eWLM) as "unique because it learns and improves in real time. Based on software algorithms from IBM Research and the mainframe, eWLM continually improves performance across a group of servers as it learns, for example, Internet traffic and applications usage patterns ... eWLM is also powerful because it (works on) servers running Unix, Linux, and Windows operating systems."

That first part of the quote above is rather important because it defines what sets eWLM apart: it can learn. Part of learning process is the acquisition of knowledge. EWLM also learns how and when to apply the knowledge it has acquired. To do that it first needs to retain a data sample of sufficient size to give it enough "experience" to know how information should be applied. More importantly, eWLM doesn't stop learning. That means eWLM accumulates an ever-increasing amount of data, resulting in an ever-rising level of processing power to compute the results and formulate its reactions. On the surface, that implies more storage and processors will be needed over time to accommodate the software meant to manage the system. But wait, there's more.

After IBM's recent eWLM demonstration, an IBM Project eLiza global executive quipped, "Project eLiza moves from self-managing technologies that reduce administration and improve the performance and reliability of individual servers to an end-to-end approach that reflects a real-world environment based on Internet and multi-tiered architectures."

That's more than a mouthful, but while pondering that consider something else that IBM is investing in: grid computing. Its basic premise revolves around borrowing the hardware power you need from where it's not in use without necessarily paying (at least the full price) for that hardware. It sounds like an ideal way of letting Project eLiza expand as much as it needs to without redlining your hardware budget in the process. Go one step further and it even sounds as if the Project eLiza components could exist "out there" in the grid and simply reach into your system to manage it with minimal local overhead.

In that odd way in which circumstances sometimes push things into a loop, we return to science fiction. This time it's in the guise of Jean-Luc Godard's 1965 film, Alphaville, the tale of a society controlled by a supercomputer, the Alpha60, "where the trains run on time but love and poetry are forbidden." It's difficult to suggest such a cynical outcome for IBM's marriage of artificial intelligence and grid computing, but if you ever encounter a hard-boiled private eye named Lemmy Caution, you might want to reconsider your plans to globalise your network.

Bill O'Brien is a freelance writer and frequent contributor to CNET and ZDNet. He writes Tech Update's weekly hardware column.

To have your say online click on TalkBack and go to the ZDNet UK forums.


You have been successfully signed up. To sign up for more newsletters or to manage your account, visit the Newsletter Subscription Center.
See All
See All