DID you know that if you can cut your servers’ power consumption by one watt, you reduce power consumption at the electricity meter by as much as 2.84 watts?

This statistic was cited by Chan Chee Wah, Emerson Network Power (Malaysia) Sdn Bhd managing director for South-East Asia, when unveiling his company’s Energy Logic roadmap for electricity conservation recently.

It is based on what Emerson calls the principle of cascading power savings. When a server consumes one watt less electricity, less power has to be converted from mains electricity, and there is a ­commensurate reduction in power loss at this conversion stage or ­stages.

Meanwhile, the server dissipates less energy into the server room as heat, so the room does not get as hot. This means less power is used by the air-conditioning system to keep the room cool. These savings add up to 2.84W saved at the meter for every one watt saved at the server.

This illustrates the general principle that power consumption saved in the demand side of a datacentre — in processors, server power supplies, storage, communications gear, and other IT systems — would reduce power demand in the supply side supporting it, which consists of the uninterruptible power supply, power distribution system, cooling, lighting and building switchgear.

Chan was promoting Emerson’s vendor-neutral Energy Logic ­strategy, which advocates a 10-step process for reducing power consumption without ­compromising IT service levels.

“Emerson Network Power does not claim Energy Logic improves datacentre efficiency per watt consumed, because there is no universal metric for datacentre output,” he pointed out.

“Instead, it is about taking ­practical steps to reduce overall power consumption. Our own studies have shown that applying the Energy Logic approach to a 5,000sq ft (500sq m) datacentre can reduce energy use by as much as half without compromising IT performance or availability.”

Even if efficiency is not the prime consideration for datacentre ­managers, and Chan said he ­understands they are more concerned with living up to ­service level agreements, these measures would help free up the key physical constraints faced in datacentres: space, power, and cooling capacity.

“If implemented together, these measures could cut a datacentre’s power consumption by half and free up as much as two-thirds of its floor space. They could help reduce requirements for UPS ­(uninterruptible power supply) capacity by two-thirds and for cooling capacity by 40%,” Chan said.

Datacentre managers should start with the IT hardware and then work their way up the power supply chain. At the hardware equipment level, they should begin by ­converting to lower-power ­processors, which could save up to 10% of overall datacentre power consumption; install more efficient power supplies, which could have a similar percentage of total power consumption; and activate the power management software built into their server processors to reduce power consumption during idle periods.

Then they should go on to ­incorporate industry best practices such as installing blade servers to replace rack-mounted servers. Blade servers share common common components like cooling fans and power suplies, saving another 10% of total power draw.

Another good practice is ­implementing server virtualisation, which decouples hardware from software so as to optimise server utilisation. A third good practice is implementing best practices in datacentre cooling, like optmising airflow, implementing hot-air and cold-air aisles, and closing up any potential air leaks.

The third phase is the ­optimisation of the IT infrastructure. This included rationalising AC power distribution to reducing needless step-downs and resulting power loss; implementing variable-capacity cooling systems that provide only as much cooling power as is needed at any given time; installing high-density ­supplemental cooling, which could cut cooling costs by as much as 30%.

Finally, system-level monitoring and control of the IT infrastructure should squeeze out another 1% saving.

“These 10 steps work best if taken together. They form a useful checklist for the datacentre manager, who can go to a consultant and use them to work up a proposal for his chief financial officer,” Chan said.

“Even though we would seem to be going against our own interests by advocating these measures — we’d certainly be able to sell more and higher capacity power ­distribution and cooling equipment otherwise — we have a role to play by educating IT managers on how to get the most out of what they do buy.

We also recognise that companies want to look for ways to reduce the environmental impact and be good corporate citizens, and this will definitely help in that effort,” he added.