For many years, energy efficiency has had a low priority in many IT shops, Jack Pouchet, director of energy initiatives at Emerson Network Power, told a crowded LinuxWorld Expo session titled "A Road Map for Reducing Energy Consumption in the Data Center." The top priority in most IT shops is delivering on service-level agreements, following by issues like performance (providing adequate compute capacity), reliability (redundancy at all steps), and security.IT cares about energy efficiency, Pouchet said, but not if it affects performance and reliability. "Many IT shops are all for freeing up power and cooling capacity, as long as it does not impact performance and reliability."

In the absence of a universal metric for data center output (like MPG for fuel efficiency), Pouchet said that discussing data center efficiency isn't meaningful. "We can only talk about reducing data center energy consumption."

The company Pouchet works for, Emerson Network Power, modeled energy consumption for a typical 5,000-square-foot data center and analyzed how energy is used within the facility. Energy use was categorized as either "demand side" or "supply side." Demand-side systems are the servers, storage, communications, and other IT systems that support the business. Supply-side systems exist to support the demand side. In his analysis, demand-side systems -- which include processors, server power supplies, other server components, storage, and communication equipment -- account for 52% of total consumption. Supply-side systems include the UPS, power distribution, cooling, lighting, and building switchgear, and account for 48% of consumption.

The distinction between demand and supply power consumption is useful because reductions in demand-side energy use cascade through the supply side. In Pouchet's 5,000-square-foot data center example, a 1-watt reduction at the server-component level (processor, memory, hard disk, etc.) results in an additional 1.84-watt savings in the power supply, power distribution system, UPS system, cooling system, etc. Consequently, every watt of savings that can be achieved on the processor level creates approximately 2.84 watts of savings for the overall facility.

Using the model of a 5,000-square-foot data center that consumes 1,127 kW of power, Pouchet then walked through an approach that would combine to produce a 585-kW reduction in energy use. Pouchet also showed the payback period of various energy-saving strategies: