October 14, 2003, 11:17 AM — The recent wave of announcements around flexible-computing models (e.g., adaptive enterprise, on demand, grid computing) promises a better future for IT organizations (ITOs), aligning IT with business-unit computing requirements. With this concept ultimately headed for airline magazines, ITOs must develop a complete understanding of the relevant economics before discussing the issue with business units. Given the general lack of maturity around understanding costs and drivers, ITOs that behave reactively will find themselves paying higher costs after adopting these new initiatives, increasing the potential for outsourcing and further exacerbating (rather than alleviating) frustrations about dealing with the ITO.
Although flexible-computing models promise an easy answer to estimating and managing demand, they set two traps for the unwary CIO. First, their convenience comes at a premium price, requiring a minimum base-capacity installation that the client must pay regardless of whether it ever uses any of the resources. Converting to a flexible-computing model also incurs other large costs (e.g., requirement that customers purchase new hardware and associated software) that customers must negotiate carefully.
Second, flexible-computing business and technical models are an intermediary between traditional IT departments and outsourcing. Consequently, CIOs who lean too heavily on flexible computing (and any associated premium) may find senior management talking to outsourcers that provide similar flexibility at a reduced cost, thereby begging the question, "Why should we have an internal IT department?"
Given the premium that vendors charge for flexible computing, the typical current model is to cover base demand for an IT system with traditional hardware purchases and software licensing, with a flexible component added on top of the base to handle demand peaks. The problem is that many IT departments have a poor internal understanding of demand economics. They do not know what their base demand is, and they are often reluctant to do the difficult work of measuring it. These ITOs will likely fall into the trap of turning all their IT contracts into flexible agreements. Exacerbating this problem, senior management may be willing to go along with these new models because of the potential accounting advantages from turning fixed capital costs into variable-expense items. In the long run, however, this strategy can come back to haunt the IT group. Near-term best practice will likely include mapping capacity requirements against long-range demand on a combined scale with declining price/performance (for both total purchase and Moore's law).