New computer servers are more powerful than their predecessors, but waste more energy, making it difficult to operate systems efficiently and without overheating chips that are becoming increasingly susceptible to hot temperatures, a Forrester Research analyst said Thursday.

Leakage -- the amount of power flowing when a transistor is turned off -- accounts for between 18 percent and 29 percent of total power usage in Intel Xeon processors, Forrester Vice President Jean-Pierre Garbani said in a Webinar for IT professionals hosted by enterprise management vendor BMC Software.

Faster switching produces more power consumption and more heat, yet the increasingly fine features of new chips must be protected by lowering the maximum operating temperatures, raising a conundrum for IT departments.

"If you have an increase in computing demand for business requirements, you're going to (quickly) come to a point where there won't be enough space in your data center. You can also run out of cooling capacity," Garbani says.

A slideshow of the presentation, titled Global Warning: Optimization Strategies to Solve the IT Energy Crisis, is available here .

The data center, Garbani says, is where business needs and the laws of physics collide. He recommends several steps for businesses, the first being to select efficient processors. He discussed three servers in the Webinar, which he says provide the best compromise between heat generation and performance: the Sun Fire T1000, the IBM p520, and the IBM x346. The latter uses an Intel Xeon processor. Of the three, Garbani says Sun Fire T1000 is the most efficient.

Second, Garbani recommends improving airflow between racks in the data center by alternating hot and cold aisles (see slide 18 in the Webinar slideshow).

He also recommends better planning data, specifically with a federated configuration management database (CMDB), which provides a single representation of data for all IT processes.

A March 2006 report from Forrester Research said management of heat in data centers became a major challenge in the previous two years, despite the crash of the dot-com economy, which might have produced immense Internet data centers.

"The problems associated with data center heat and power management have come roaring back over the past two years, thanks to complex interactions between multiple factors, ranging from chip technology to data center design," the report states. "A combination of rising processing requirements, potentially increased demand for rapid scaling of processing capacity, energy costs, and the politics of conservation have also driven power and cooling onto IT's priority list."

David Wagner, director of capacity management and provisioning solutions at BMC, said at the Webinar that companies must closely analyze the number of servers and how they are used.

"The right answer is to make sure you have the right number of servers and the right power of servers. It really gets back to capacity management," Wagner says.

BMC says its capacity management and provisioning software supports functions of an ITIL, or IT Infrastructure Library, including monitoring, analysis, tuning and implementation.