Tech Industry

Happy talk and the realities of IT

Dynamic, adaptive, on-demand IT may be a fabulous goal, but Illuminata founder Jonathan Eunice says the fact is that technology doesn't--and won't always--work as advertised.

If there is a major IT company out there not currently promising to deliver a new era of peace and harmony via a hyperefficient, ultra-dynamic, instantly adaptable data center, I am unaware of it. I am inundated with grand visions and "initiatives of industrywide importance."

Sun Microsystems nudges us toward N1; Hewlett-Packard advocates the Adaptive Enterprise; IBM opines that all organizations should be On Demand.

Have you ever heard such a flood of happy talk in your life?

If you're in IT, you almost certainly have. Our industry is at it again: Virtualized, autonomically adaptive, on-demand data centers--whatever all that really means-?are the anointed next big thing. So far, there's more motherhood and apple pie in the pitches than there is any progress. There are a few early stage products, but most of the visions are just "warm fuzzies" and blanket reassurances. Who, after all, could possibly be against an adaptive enterprise?

But it's hard not to notice that today's data centers rarely, if ever, operate this way. If glorious harmony and adaptability are so near at hand, why don't we witness at least a modest helping of it in today's environment?

First, the technology doesn't always--and won't always--work as advertised. What makes sense in a lab setting, or even at a beta test site, doesn't always work efficiently in the haphazard mesh of corporate data infrastructures. Just as important, it's not just a technology problem.

The mechanisms that enable change aren't free-?and the cost isn't just monetary. As in nature, most flexible and resilient organisms aren't necessarily the largest, the fastest, or the cutest. We'll have to forgo other opportunities and broadly reprioritize IT investments to favor flexibility.

For example, departments, divisions and lines of business will have to aggregate their resources. If resources are going to be reassigned to different jobs on the fly, they're going to have to be reallocated from somewhere. That means either buying a lot more gear in the first place?-a suggestion that gets more bitter laughs than sober consideration in today's economic environment?-or combining the gear of several different business units so the total mass can be allocated and used more efficiently. That means sharing, a great concept but one frequently unpopular in practice.

If you want to see the economies of scale promised by dynamic data centers and grids, you have to sacrifice autonomy that has grown to almost ludicrous levels in some IT fiefdoms, and actually share resources with your peers in other groups. Even units that have jealously guarded their own prerogatives for many years will have to pitch in and coordinate, or it won't work, no matter how good the technology is.

So far there's more motherhood and apple pie in the pitches than there is any progress.
Sharing also has secondary, but inexorable, implications. Like the need for a common vocabulary in which to discuss, and measure, IT outcomes. Geek metrics such as server uptime are kaput; providing IT as a dynamic service requires application and service-oriented management-?a shift that must apply to whomever's involved in IT planning, purchasing, development and support.

All this means cultural change?-and a big helping of it. If there's anything people hate more than sharing, it's change. It'd be easy if it were just bringing in a new rack of blade servers, a grid toolkit, or a network switch. No such luck. There's no simple appliance you can install, and (as the IBM ad says) no magic pixie dust you can sprinkle. Going from static to dynamic requires greater organizational fitness and poise. The real goal, after all, isn't having an agile data center. The real goal is having an agile enterprise.

That demands strong motivation and leadership; it will also take time and effort. As a result, not every enterprise will be able to make the transition. A number will lack the energy and enthusiasm to make the turn.
If there's anything people hate more than sharing, it's change.
As a result, they will not only waste a lot of time and money, they'll end up being at a competitive disadvantage in a world where IT is the beating heart of commerce.

Dynamic, adaptive, on-demand IT is a fabulous goal. We're delighted that the industry has taken it up as its mission. But let's be realistic: Just saying that you're going to inject orders of magnitude more intelligence, flexibility and harmoniousness into an IT structure is not the same thing as doing it. Sometimes you have to inject the flexibility and intelligence elsewhere first. There's an awful lot of change required before we get to this promised land, and only some of us will be making the trip.