Processor Magazine – February 23, 2007

Is it hot in here, or is it just us? As blade servers and high-powered storage become more common, data centers are warming up—enough to make the president take notice. Recently, President George W. Bush signed legislation asking the EPA to study how data centers use energy. Others, including vendors of cooling technologies, have called the problem a crisis. But is it?

“I would say that it is a crisis for some customers,” says Mike Hagan, vice president of Lee Technologies and an expert in data center design. “It’s becoming a major concern for most,” he adds.

It’s also becoming a source of innovation, according to Lisa Raffin, vice president of professional services with VFA, a leading facilities management firm. Raffin believes the problem is creeping up to crisis levels, but its solution could leave us with everything from better, renewable sources of energy to smarter, more efficient servers. In IT, crises often become catalysts for invention. “History has shown that these are not false alarms,” says Raffin, “and that we will make significant advances in building technologies as a result.”

What’s Causing It?

"The outlook for the next few years is a warm one, in part because Dell, HP, IBM, and others have invested billions to develop ultra-dense blades and expect a return on that investment. “They need to sell a lot of blade technology,” says Blake McLane, vice president of strategic business development at colocation provider CyrusOne. He notes that server manufacturers can push the adoption of blades by cutting support for older models.

But McLane also believes that blades and business cycles aren’t the only problem, or even the root problem. Indeed, newer systems consume less power and throw less heat than older, equally dense models. And most data centers have enough cool air to chill things out."

So what’s the problem? Airflow. Like a damaged heart, most data centers suffer from poor circulation; the cool air they have is dammed or pooled or badly distributed. One solution is modular cooling that can be configured—or reconfigured—to match a data center’s changing needs, much as CyrusOne does in existing centers and plans to do in a new 60,000-square-foot data center to be built in Houston.

“We’ve designed the power and cooling structures in a very modular way,” says McLane. And while CyrusOne’s center is a Tier 4 facility, even smaller data centers can benefit from its approach, largely because they’ve grown in what McLane calls “an evolutionary way.” Over time, they replace old servers with new or add servers as computing power is needed. Thus older, less dense systems exist side by side with clusters of highly dense machines that result in hot spots.

With a modular system, CyrusOne can push cold air to the places that need it. “We’re able to concentrate the distribution of airflow and CFM specifically to the zones that require it,” says McLane.

The Right Planning.

Hagan of Lee Technologies says the heat problem leads data center managers to pay for solutions they may not need. “It’s very easy to become enamored with high-end technologies and software and applications. They’re flashy,” he says.

But preventive maintenance on your current cooling equipment, a distinctly unsexy job that’s often ignored, may give you enough cool air to cope with the servers and storage you have now. Even better is careful planning, especially if you’re building from scratch.

“When you’re in a new environment, if you’re able to build a deep enough floor plenum, keep floor obstructions to a minimum, take advantage of hot aisles and cold aisles, and provide the proper cooling structure, you can meet pretty rigorous needs with what I would call conventional cooling methodologies,” says Hagan. “The challenge comes when you’re looking at existing clients.”

If you can’t build from scratch and you’re coping with high heat and higher cooling bills, an approach like the one taken by VFA might be useful. “The first step is to establish a baseline of the current condition,” says VFA’s Raffin. Finding a baseline involves a cooling audit to determine everything from capacity to latent and sensible loads, which “helps identify design vs. equipment issues, as well as other energy savings opportunities,” she adds.

Next up? A plan for the future. Raffin believes it’s every bit as important as looking at current problems because there are few business needs in today’s high-tech landscape that won’t demand computing power, whether it’s a week, a month, or a year from now. And whether you’ll need servers for 10 users or 10,000, there’s no time like the present to start planning.

Planning, Talking & Planning To Talk

Planning for a cooler future is not as easy as it seems. All too often, companies fail to estimate how much cooling they’ll really need and presume they’ll get five to seven years out of a cooling system that only covers their needs for the next 24 months.

“Not having a crystal ball and not knowing what to invest, the alternative is you overbuild and don’t use it,” says Blake McLane of colocation provider CyrusOne. “And your financial folks don’t like you taking all that depreciation every year when you’re only running at 30 or 40% capacity,” he adds.

Better planning can help—providing the lines of communication are kept open. “IT personnel are quickly adding high-density equipment without having a solid discussion with their facilities management people,” says Mike Hagan of Lee Technologies. And when the IT staff adds equipment without keeping the facilities group in the loop, problems are sure to appear.

This article was originally published online at Processor Magazine

Copyright © 2006 Processor Magazine,