Optimize, one of InformationWeek's sister publications, recently ran a very good article by two consultants that I urge everyone in IT to read. It's about how to prepare for utility computing, and it contains some sage advice about steps to take. Their recommendations include starting in relatively small areas that already have homogenous environments (in other words, single-vendor applications or something close), implementing a SAN and chargeback system, and other things.

Even more interesting is that the article raises an intriguing question about why the vast majority of IT organizations are holding back from investing in this computing model, at least so far.After all, the notion promises resources-on-demand, an environment where you can get what you need to meet peak needs while paying for only what you actually use. One server that's underused at any given time can kick in to power the CRM system, for instance, during major usage periods, and so on. It's kind of like electricity--hence the term "utility computing."

As an aside, perhaps one issue is some pushback, however unconscious, about the term itself. If we're heading toward "utility computing," does it mean what we're doing now is "useless computing"?

But I digress.

The consultants--Jim Cassell and Bruce Guptill--have some theories about why ITers have been reluctant to move to the utility model, though they maintain that's about to change big-time. (Full disclosure: A long time ago, when I was a reporter at another trade publication and Jim was at another consulting firm, I used to bug him on a regular basis and pick his brain. Kind soul that he is, he would let me. He was also the source of that great line about why Wintel-based servers couldn't step in for mainframes as the backbone computing platforms in most big companies, at least at the time we were regular correspondents in the early 1990s: "You can't replace one ox with a thousand chickens.")

To help explain the problems around utility computing that most companies face, the writers go through a list of eight problem areas. Most relate to organizational issues, including how to get a bunch of business units that may be operating independently to agree to share the central IT "resource."

I could be wrong, but I believe that this buy-in and the resulting ceding of control are probably the biggest roadblocks to adopting utility computing en masse. Most large and medium-sized companies have gone through the yin-yang pull of decentralizing and recentralizing IT numerous times. Those that are in the latter camp have usually wound up there either because of business reasons--multiple acquisitions and mergers with incompatible IT infrastructures--or the business units had really bad response from the central IT group and struck out on their own. ("We'll get our own IT department," etc.)

So there's that and all the other issues raised in the article, and one more biggie: The technology isn't really "done" yet, not to be able to deploy utility computing in any serious enterprise-wide way. Sure, there are small rollouts that can be done--and they are--and it's probably a great idea to get some experience at this while it's still small and manageable and when your entire infrastructure won't be blown away by any glitches.

Fundamentally, though, the "MEGO" (my eyes glaze over) factor is huge here--and by "here" I mean utility computing, autonomic systems, and pretty much next-generation anything. After decades of hearing about how technology is going to make their lives better, their infrastructures more manageable and/or less expensive, their smiles brighter (or just pick your benefit), I think IT professionals are casting an extremely wary eye on just about everything right now. CIOs are, and for very good reason, skeptical of any promises because they're tired of getting beaten up by their business management.

I have no doubt that we will someday get to utility computing and the rest. Like just about everything else in the industry, though, it's going to take a lot longer than some may hope.

What do you think? Add your comments below.Fundamentally, though, the "MEGO" (my eyes glaze over) factor is huge here--and by "here" I mean utility computing, autonomic systems, and pretty much next-generation anything. After decades of hearing about how technology is going to make their lives better, their infrastructures more manageable and/or less expensive, their smiles brighter (or just pick your benefit), I think IT professionals are casting an extremely wary eye on just about everything right now.