Build your data center here: The most energy-efficient locations

Computerworld - It's almost a no-brainer that bucolic Quincy, Wash., would be a more environmentally friendly site for data centers than, say, New York is. But according to a new study, Chicago beats all 20 of the other locations that were looked at -- even Quincy -- in terms of a data center's carbon output.

On the other hand, going by the cost of electricity alone, Westport, Pa., and Quincy come out on top. But cities such as Miami and Atlanta have it all over the likes of San Jose and Boston.

So say Base Partners Inc., a data center consultancy, and Glumac, an engineering services firm. The two companies, both based in San Francisco, teamed up to study the "green" credentials of 21 locales that already have pockets of data centers. The locations studied included New York, Chicago, Phoenix, El Segundo, Calif., Ashburn, Va., and Quincy.

"As we talk with enterprise users, we're seeing a lot of pressure on costs and a new focus on internal sustainability programs," says Aaron Wangenheim, president of Base Partners. "The data center is one of the biggest environmental offenders, because it's one of the largest users of electricity. Our clients want to know what they can do to save money and do something good for the environment that's not going to cost a lot."

To help figure that out, Glumac and Base Partners devised a fictional 135,000-square-foot data center with a raised-floor area of 110,000 square feet. That's a bit different from the typical data center space allotment, which generally tends toward a 2:1 ratio -- in other words, out of every 1,000 square feet of total floor space, 500 are for a raised floor.

The made-up data center is also on the large side; corporate data centers generally range from about 30,000 square feet for a midsize company to upward of 300,000 square feet for a bank in a large city, according to Michael Steinmann, managing principal at Glumac's San Francisco office.

In the made-up example, a constant power density of 150 watts per square foot was assumed. And Bergman says that in cases where the locations offered a choice of electricity providers, he did what most customers do: He chose one.

One factor that wasn't considered: the efficiency of a data center's cooling and heating systems. "So many factors impact the performance, and there are many different types," explains Aryn Bergman, a mechanical designer at Glumac who ran most of the mathematical models as part of the study. "It also depends on the climate that the mechanical system is used in. We would have had to include a number so vastly oversimplified to be useless."

One way to figure this for your own data center calculations, Steinmann says, is to assume that half again of the overall power consumed by servers is needed by their cooling systems. So if you have 200 watts per square foot of power going to the raised floor, assume that another 100 watts per square foot will be needed for the mechanical system.

Given all these variables, Chicago and Quincy came out way ahead of all the other sites studied in terms of annual carbon production (see Figure 1). Bergman says he considered how much power the local utility companies generate via the use of oil, natural gas, coal or renewable sources, including hydroelectricity, solar energy, wind and biofuels.


Figure 1. (Source: Base Partners/Glumac)