Promises abound when it comes to utility computing, but what about the practicalities and the politics?

It took a while for the geologists to accept one of the side effects of utility computing. Arriving at work for an early start, the geologists discovered that although they were ready to work, their computers were not.

The geologists' oil company employers had implemented a utility computing model, which meant the big, powerful machines sitting on the geologists' desks were plugged into a grid and exploited to perform other processing outside "normal" business hours. What the utility computing schedulers had not taken into account was that "normal" business hours and oil company geologists' working hours might not align.

What the geologists found was that although the promise of utility computing has been well canvassed (see "IBM's New Hook", CIO July and "Plug and Play", CIO May) the practicalities and the politics also need careful scrutiny.

There have been suggestions of 50 per cent reduction in operating costs for those organisations that implement utility computing well. Companies such as American Express, AXA, Deutsche Bank, Boeing, BP, Ford and Daimler Chrysler are all seeking big savings from utility computing trials. Some are experimenting with their existing infrastructure and attempting to transform that into a utility for the corporation, which can then be costed as such. Others, such as AXA, are going outside - in AXA's case signing a $US1 billion deal with IBM to provide a utility computing service.

Not surprisingly, CEOs and board members read statistics and blue chip names like these, splutter their coffee, and demand an explanation from the CIO: "Why aren't we implementing utility computing?"

While dabbing at coffee stains, the CIO might well remind the CEO that in the phrase "utility computing" the word "utility" is an adjective, not a noun, and that the definition of utility is: "provided, designed, bred or made for usefulness or profitability rather than beauty". Also, in order to liberate that inherent usefulness or profitability, often a business will need to change substantially its corporate culture, reorganise the workflow and business process, and tinker with the structure of organisational responsibility.

It should at least buy some time.

Analysts writing in the McKinsey Quarterly earlier this year predicted that companies that only slightly reorganise to account for utility computing will liberate infrastructure savings of 15-18 per cent, whereas organisations that fully overhaul their processes and procedures can expect savings of up to 30 per cent. A firm saving 15 per cent is doing well, but remains at a competitive disadvantage compared to the rival snaring 30 per cent savings, so understanding what reorganisation might be beneficial is important.

The degree to which an organisation has to reconstruct itself in order to liberate any savings will depend on which variant of utility computing it adopts. For example, it could buy in computing from a third party in an outsourced utility arrangement (the AXA model), or it could act as its own utility. With the latter option, when a company is pooling its international computing infrastructure, which can then be made available to business units as it is required, there is a need for reorganisation in terms of management responsibilities and process, both at the business level and within the IT department. Where the utility computing is provided by a third party there will be a series of other issues to contend with, many of which will be handled by lawyers, such as the requirement for significant due diligence about the capability and stability of the service provider, the requirement potentially for key software and data to be held in escrow and the creation of robust service level agreements.