Utility Computing Meets Real LifeUtility Computing Meets Real Life

Sun has many of the pieces required to deliver a utility-computing package, but it's unclear how well the company has integrated provisioning software acquired last year with Terraspring Inc. and storage virtualization acquired last year with Pirus Networks. In addition, "Sun has gaps against IBM and HP," because it lacks management software on the level of HP OpenView and IBM Tivoli, Gillett says. Another challenge is proving that N1 can manage heterogeneous product environments.

Nielsen Media Research, a user of Sun servers and storage, likes the idea behind N1. "If they've established that as a corporate vision and they succeed at that, it's all the right stuff," says Kim Ross, CIO of Nielsen, a subsidiary of $4 billion-a-year Dutch business-information and market-research company VNU N.V. "The end goal is to port Nielsen's complex computing model to one where components are virtual and the company can look at these resources as a pool."

Like many companies, Nielsen Media has a lot of servers and storage provisioned on a per-project basis. "That leads to a lot of waste," Ross says. And it gets worse each time Nielsen adds a server dedicated to a specific app. "We'd love to be able to share resources once the technology becomes implementable in our environment."

Still, Sun's Java 2 Enterprise Edition programming encourages IT departments to separate technology objectives into tiers, Ross says. As a result, apps written in J2EE aren't always easy to integrate on the same multiprocessor server, so they each get a dedicated server. "That means the provisioning of resources is harder because you're managing a greater number of servers," Ross says, noting that Nielsen has different servers doing statistical analysis, interfacing with clients, and providing database access.

In the end, the promise of less waste, higher utilization, and cost savings will spur the adoption of utility computing, Ross says. What's less clear is when the technology will gain critical mass in the marketplace. "Right now, it's just a strategic vision for Sun. I take it for that."

Attempting to gauge the size of today's utility-computing market is tough. "We're not only talking about a combination of software, servers, networking, and storage," Gillett says. "We're also talking about an overhaul of those markets."

HP acknowledges that its Adaptive Infrastructure strategy will take some time to find footing at a time when companies aren't sure whether to spend or save. What's certain in HP's eyes is the need for utility computing. The company says that the average data center uses only about 35% of resources at any given time. "Utility computing is more of a vision in the marketplace and a vision within HP," says Nick van der Zweep, HP's director of utility computing. "Someday server, storage, networking, and appliances will be interconnected."

The challenge for HP, IBM, Sun, and others is to sell a new computing paradigm to a skeptical market. It takes a lot of work to turn an existing IT infrastructure into a utility-based infrastructure, says Andrew Schroepfer, president of Tier 1 Research. Many companies, particularly small and midsize businesses, are likely to outsource to achieve the short-term cost-savings that utility computing offers, he says. Large outsourcing companies such as IBM and HP are simply offering their customers a utility pricing model, since IBM and HP aren't yet running their own internal IT operations as utilities.

"Service providers can meter usage and bill based upon how much resources are being used," Schroepfer says. "But in terms of creating their own big utility data center, no one's done that yet."

Companies don't have to implement all of the available utility-computing technology to see benefits, Gillett says. They will probably implement the technology in a piecemeal fashion as the time comes to replace elements of their IT environments. "We don't see a lot of customers doing this across the board."