How to Increase the Reliability of Your IT Infrastructure Using Predictive Analytics REGISTER >
Virtualization won't solve all your problems. The virtual servers you create are still IT assets that need to be administered and managed, as Qualcomm's Fjeldheim is quick to point out. Frank Gillett, a principal analyst with Forrester Inc., agrees. "If you put ten virtual servers on a single machine, you eliminate nine pieces of hardware, but you still have ten operating systems to maintain." John Hinkle, vice president and CIO of Trans World Entertainment Corp., has also found that "managing the flexibility [provided by virtualization] is another skill set that we needed to develop and train people in."
Over time, the various virtualization products currently needed to handle virtualization for storage, servers, networks and other infrastructure elements are likely to converge into a single administration tool. Products such as IBM Corp.'s newly released Virtualization Engine 2.0 are meant to make it easier to virtualize across technology layers and manage it all through a single console. Still, getting all those virtualization products to work together remains difficult.
Meanwhile, certain computing tasks will always demand a dedicated machine: "If you have to squeeze the most performance out of a piece of hardware, then you still need the physical hardware," says Adams at J.R. Simplot.
Perhaps the biggest issue IT organizations face, according to a recent report from Gartner, is that there is no "defined and acceptable set of metrics and pricing mechanisms to support virtualized, on-demand services. Almost everything in the IT infrastructure industry has a cost or price associated with a physical box." It will take years for IT vendors to overhaul their pricing to reflect the flexibility and efficiency inherent in virtualization and related technologies. And IT organizations will likewise have to rethink how they charge their own customers for the IT resources they consume.