Utility computing holds the great promise of being able to more efficiently harness often-underutilized corporate IT infrastructures -- a big selling point for today's cash-strapped CIOs, who would be happy to channel part of their annual spending into projects that could boost revenue.
But the promise is not without pain. The technical issues involved with utility computing are complex enough (see part one of this series,
Implementing such a strategy means overhauling the IT department, just for starters. It also involves the incredibly difficult task of getting departments and functions to share computing resources and, most important, making sure that IT functions are properly mapped to business processes.
If all this has a familiar ring to it, there's good reason. "At the end of the day, CIOs and CTOs understand the technology issues clearly," says Jamie Gruener, a senior analyst at the Yankee Group, a Boston-based research company. "The missing link is how to link technology issues to business processes. That has been a 35- to 40-year discussion, and we're having it again."
If utility computing is to work, a bridge must be built between a company's IT infrastructure and its business units. There are no agreed-upon best practices or common methodologies at play yet, although many consulting groups are busily building their utility practices. (IBM, for example, likes to present its Global Services unit as a leading practitioner.) But experts like Gruener say that there are several key areas that CIOs should address in order to turn the odds in their favor when it comes to utility computing.
Choose a utility model that fits business needs
"Most people think that implementing utility computing has to be all or nothing," says Jeffrey Kaplan, managing director of ThinkStrategies, in Wellesley, Mass. "But it doesn't have to be that way." He advises CIOs to take a look at their business functions first, then evaluate utility options based on that. While some companies may indeed opt to build a large utility model in house and others will choose to outsource to utility service providers, Kaplan says that most will end up with hybrid models. The choice, however, should be based on what serves the business best. "Start by deciding what's core and non-core to the business and therefore the IT strategy," says Mike Riegel, manager of the global marketing management team at IBM Global Services, in Armonk, N.Y. By starting small, a CIO will have a solid record of success from which to attempt more strategic utility projects.
Craft service-level agreements based on business policies
It's vital that business executives know what to expect when they sign up for utility computing, so it's vital to have a service-level agreement -- whether it comes from an outsourcer or is an internal IT agreement -- that delineates exactly what will be delivered. Even more important, those SLAs need to be based on realistic business priorities. Kaplan says that giving business units some sort of management software that lets them document performance can be a big selling point, since it allows them some sense of continued ownership. "Customers often want the ability to monitor the computing power or service levels," he says.
Set resource priorities
Fact of life: Some data is just more important than other data, and the CIO needs to figure out the food chain before implementing utility computing. Why? Utility computing, with its dynamic allocation of resources, will only succeed if there are clear guidelines about which applications get which resources. "For example, think about granting server priorities," says David Kelly, principal at Upside Research in Newton, Mass. "Is payroll more important than order processing? It requires strong CIO leadership to create a private internal utility."
For example, Riegel says that, at Charles Schwab, the company decided on different levels of service based upon preset business policies. "So platinum traders get much faster response times than small individual investors," he says. "It's a very tangible link between utility computing and driving more value to core competencies -- in this case, it's the customer relationships."
Plan for IT reorganization
When the CIO decides to transform his infrastructure to a usage-based model, he also needs to revamp the IT department to support the new architecture. "CIOs will need to make IT workers more versatile, so people can be moved around as requirements dictate," says Corey Ferengul, vice president at Stamford, Conn.-based Meta Group.
The traditional silo approach to IT, in which workers remain in function-specific areas, such as servers, storage, and databases, won't work in a utility computing model, says Upside's Kelly. "In the utility model, they're all closely interlinked, so you can't have those silos," he says. "IS organizations will need more horizontal instead of vertical processes. Resources must be interlinked, and people have to be able to talk to each other and understand what business needs are going to drive the consumption of those resources."
Thomas Bittman, a vice president at Stamford, Conn.,-based Gartner Inc., says companies should create what he calls "dotted line organizations" based on services. "You want the services to be king instead of the architecture," he says.
Don't forget business process modeling
Last but most certainly not least, CIOs need to tackle the enormous task of mapping important business processes to the new infrastructure, to ensure correct resource provisioning and chargeback services. This means an intense interaction with IT and the line-of-business managers, Bittman says.
"A whole new language needs to be created between business and IT," he says. "How much resources are given to each application needs to be based on business policies."
Ferengul agrees: "If you don't have well-documented and flexible business processes, you'll have a hard time making use of flexible IT processes," he points out. And that means that CIOs need to understand very clearly how business processes link to the IT resources that support them, says Kelly. "You need to integrate the process [with] the technology to have the linkages clearly understood," he says.
None of this will be easy; CIOs have been struggling with some of these tasks for years. But as the drumbeat of "doing more with less" continues to roll across the land, it may well be that utility computing is the idea that finally drives these tasks to completion.
This was first published in May 2003