Other issues to navigate include determining which business unit has priority access to the computing power, and when. For example, some business processes may be scheduled overnight when a lower processing rate may be available (hopefully without upsetting the geologists). Scheduling will take on a very sophisticated complexion and could become extremely political - with business units fighting each other for processor time - without deft handling. Clearly this is a role that would have to be overseen by an individual who could identify which information processing tasks affected particular workflow, and the cascading effects of workflow scheduling. For the enterprise that is of sufficient scale to consider becoming its own utility, there is the additional technical burden of developing and managing the pooled infrastructure and apportioning costs.

Alastair McAulay, a senior consultant in the London office of PA Consulting, has recently prepared a paper on the topic of utility computing. McAulay believes that "utility computing is now at roughly the same stage of development as the Internet was in 1994"; but he also says that "the inherent contradiction with utility computing is that complex and distributed IT infrastructures are combined to provide a flexible easy-to-use service". It will not take much guesswork from CIOs to work out who will be held accountable for overcoming the contradiction. According to PA, technologies are emerging designed to allow automatic configuration and maintenance of utility grids, but these are in their infancy and incomplete.

If the technology challenges seem daunting, Alastair McAulay observes that "utility computing isn't just about technology, it is really about changing service provision, relinquishing control of IT resources and working in new ways to take advantage of the opportunities. All of this is going to be daunting for any organisation".

It probably explains why "so much said, so little done" - as IDC's chief services analyst Merv Langby says of utility computing.

"The utility model is supposed to be a solution for enterprises to access technology, skills, hardware, software, tools and networking in an environment that needs to be more dynamically responsive to constituents," Langby says. "In theory, CIOs should have less to worry about as far as overcommitting their company to IT resources is concerned because this is supposed to put the power in their hands; but there will still be the classic concerns such as the psychological aspects of control, and security and the confidentiality of information."

Easier Said Than Done

Allan Granger is a partner and has responsibility for IT at accounting firm BDO, offering him the perspective both of CEO and CIO. With 27 partners and a staff of 200 in Sydney, Granger is attracted to the idea of paying a daily or hourly fee for information technology services, especially when such an arrangement might afford the company access to tools and applications that it cannot currently justify buying.

But his CIO brain tells him: "It's not that simple."

One complication for companies of BDO's size, which would have to find a utility provider rather than act as their own utility, is that the user would be "tied a little bit tighter" to the supplier. "If your supplier went belly-up . . . clearly then you have issues of how you get data from one provider to another," he says. "It's like any contract you go into. These issues would all come into the business decision and that includes continuity of service and what would happen if a supplier failed." Issues of data and software escrow would have to be explored in order for a company to be assured that it could continue to operate even if its utility computer provider did not.