Jan 06 2012

Tweaking the Temperature in the Data Center

Reduce inefficiencies using these power and cooling management practices.

Climate change may be a hot topic of conversation these days, but only in IT circles might a discussion about global warming incite sweating over the temperature of the data center. School IT leaders can either ­dedicate a significant portion of their budget to power and cooling or risk downtime if and when their hardware overheats and fails. The trick is avoiding such failures without overspending on technology to mitigate the threat.

Uptime Institute research shows that data center power consumption has increased more than 600 percent in just seven years, and that power and cooling accounts for anywhere from 50 to 70 percent of today’s data center costs. Its inaugural industry survey, conducted in spring 2011, is equally telling: 36 percent of the data center owners and operators who participated reported that their facility would run out of power, cooling and/or space this year. And the 70 percent of respondents who measure Power Usage Effectiveness reported an average reading of 1.6 to 1.99 — well above the 1.0 that The Green Grid considers ideal.

Power Plays

In 2010, IDC Research Manager Jed Scaramella reported that the energy expense associated with powering and cooling the world’s servers had increased 31.2 percent since 2005, and that server energy expenditures outpaced spending on the servers themselves during the same period.

It’s not all that surprising, then, that IT executives surveyed by IDC consistently have ranked power and cooling as their “top item of concern.” As Scaramella writes in “Improving Power and Cooling Efficiency in the Datacenter,” IT organizations’ stated initiatives may include improving ­energy efficiency, reducing or ­controlling operating expenses, or even looking to go green, but “IT availability overrides all other factors and will not be compromised to achieve other efficiencies.”

In this age of austerity, district IT leaders are looking to strike the right balance in pursuing (and achieving) all of these goals.

Real-World Approach

To overcome the power outages that frequently plague the Dallas area, Neil Bolton, network analyst for the Cedar Hill (Texas) Independent School District, implemented an APC-based power and cooling solution, which eliminated the capital expense of dedicated cooling in the data center and saved thousands of dollars. “The solution shares the building’s chiller, so we’re not wasting as much power to cool our data,” he says. “The hot-aisle containment system really quarantines the heat.”

As Bolton’s team discovered when devising its strategy, planning is key. “It’s important to really understand your existing power requirements to ensure that you can properly scale out for growth,” he cautions. “It’s easy to buy cheap and just get a solution that will work now, but you’re really shooting yourself in the foot over the long term if you do that.”

As a bonus, a carefully considered data center design can ­extend the lifecycle of all hardware, leading to fewer repairs and upgrades — and a healthier bottom line. “We needed a solution with a 10-year refresh cycle,” says Peter Poggione, IT director for the Mattawan (Mich.) Consolidated School District. Thanks to Liebert uninterruptable power supply units, “we’re on pace to make our goals,” he says.

“Don’t skimp” when it comes to maintenance, Cedar Hill’s Bolton adds. “This is the core of your operations, after all. It really helps to have a proactive maintenance approach to the whole life of the solution.”