As I See It: My Place Or Yours

by Victor Rozek

"Fashion is a form of ugliness so intolerable that we have to alter it every six months."
--Oscar Wilde

Not all that long ago, IT organizations were grappling with one of the most vexing question since the first Neanderthal pondered "how high the moon?" And like all great eternal questions, this one had no single answer that would satisfy everyone. More elusive than the meaning or life; more volatile than debating the existence of God; more daunting than the Sophie's choice of paper or plastic; the question straining the stout minds of IT managers everywhere was: centralized or distributed?

Unfortunately, it seemed that neither choice was satisfactory for very long and many companies alternated between the two, switching back and forth every few years when the problems solved by choosing one alternative created problems that could only be solved by choosing the other.

Mainframes offered high control, but low flexibility; distributed systems were elastic, but they multiplied like Starbucks and managing them was like tackling fog. But in keeping with the cyclical nature of intolerable fashions, the quandary is back, freshly laundered for the new millennium.

The question straining IT managers these days is: My Place or Yours? "My place" refers to the radical notion of actually having an IT installation on-site. You know, those old fashioned ideas like in-house control of your data, and hardware and software that you actually purchase, install, and maintain. "Your place" is the casa of one of several major IT service providers whose vision is the creation of one, quasi-virtual, holistic, and ecumenical computer center where all necessities are provided, all wants are fulfilled, and all concerns assuaged.

But like blind men groping different parts of an elephant, providers can't agree on what to call the pachyderm. IBM, which invests about $10 billion a year on the care and feeding of its elephant, first called it On Demand Computing. (Clever, because it gives you the illusion of being in control.) Sun Microsystems originally had a name for it that sounded like was a 1930s-era singing group, WebTone. (Here, for your listening pleasure, direct from our studios atop Rockefeller Center, are the WebTones . . . .). Now Sun just calls it the Sun Grid, like it owns the only one. Hewlett-Packard executives have apparently been watching too many Star Wars movies because they chose the galactically grandiose Planetary Scale Computing. (Just how does one inquire with a straight face about Planetary Computing? Ask to speak with Chewbacca?) Eventually, they changed that to Utility Data Center, and when that didn't work out, they went with an even vaguer "Adaptive Enterprise" moniker.

Ever conscious of the advantage of being the first to define fashion trends, IBM upgraded its pedestrian-sounding On Demand Computing to the equally pedestrian-sounding Utility Computing. As snappy names go, that's like changing Datsun to Nissan. "Utility" refers to marketing IT services the way we sell electricity--like a commodity that is available on demand and paid for by usage.

The godfather of Utility Computing was actually former IBM chairman and CEO, Louis Gerstner, who dumped a lot of money into IBM Global Services during his tenure and on his way out the boardroom door. The investment paid off. Since 1992, Global Services has grown from a modest $7.4 billion to $46 billion in revenues and today the division commands about 175,000 employees, or roughly half of IBM's workforce. Services proved to be so profitable that IBM followed its investment in Global Services with the $3.5 billion purchase of PwC Consulting. Armed with infrastructure and expertise, IBM now stands ready to take the technological load off the backs of its customers. They ain't heavy, they're our revenue stream.

Like a successful movie that is then imitated ad nauseam, the concept of Utility Computing spawned a number of variations. There are a half dozen trends including Grid Computing, Autonomic Computing, Adaptive Management, Metered Services, and Services on Demand, that conceptually serve the same marketing stratagem. Customers are being funneled toward an IT future of massive centralization but with a new twist: it will no longer be under customer control.

Autonomic computers will supposedly be self-managing enough to solve their own problems and can therefore reside outside our awareness. Grid Computing allows these dispersed autonomic systems to be shared and managed like one big, happy virtual system. And with Metered Services, customers can have access to unlimited IT resources, but pay only for what they use. Add a generous dash of bandwidth and sufficiently open standards to ensure interoperability, mix vigorously and poof!, no more IT department.

But without the possibility of empire building, what is to become of IT managers? No more data centers, no new toys, no control over hiring or budgets. What's a benevolent despot to do? The king does not willingly turn the castle over to the peasants. "What about data security?", they will cry. "What of payroll records, proprietary information? What about in-house control?" And they may be right, but who will listen? Not the CEOs counting the bonuses they received for downsizing IT. The reply from the 32nd floor will be, "Let them eat unemployment benefits."

The astute CEO will begin to see Utility Computing as a prelude to Utility Management. After all, with no data center to run, why dole out a full-time manager's salary with all those annoying benefits and high health care costs? Besides, how much time do managers waste shuffling from meeting to meeting, eating up precious hours but accomplishing precious little. With Utility Management, you just rouse your manager periodically when he is needed (a toe in the ribs, like waking a sleeping dog, is usually effective) and pay only for the time during which he is actually doing something useful. You can imagine the savings!

If the on-demand concept works with management, it should be applied to teenagers and telemarketers.

Marriage may solve some problems, but it creates others, and getting hitched to a single service provider will not guarantee happily ever after. Fashions, as Wilde noted, eventually become intolerable and will change. And so will Utility Computing. But not because of peak-usage brown outs, although they are amusing to ponder, and not because it doesn't offer some advantages. Utility Computing will ultimately fail because it removes the single most essential element of corporate governance: the presence of someone in-house to blame when things go wrong.

Blaming remote service providers is not nearly as satisfying. Would you rather descend through endless levels of phone mail hell, listening to recorded messages assuring you that your call is important; or would you prefer to have your hands around the neck of the offending employee? Just as I thought.

Utility Computing will create such a degree of dependency that it will be almost impossible to get rid of the service provider because there will be insufficient in-house expertise to replace it. Ultimately, it is easier to replace an IT manager who is not performing adequately than to replace an IT service provider who happens to be running your entire company. If the provider fails, the CEO fails, and CEOs are CEOs because they prefer control and keeping their tush well covered against unnecessary risk. Corporate accountability simply works more predictably with a designated fall-guy. Now, if Utility Computing vendors could provide one, then they'd really have an attractive concept.

Because, as Oscar Wilde noted: "It's not whether you win or lose, it's how you place the blame."