Hurricanes. Tornadoes. Ice storms. Rolling power outages. The list of threats facing today’s data centers is seemingly endless and serves as a continuing reminder that unprepared sites will face severe ramifications in the event of a disaster. But developing a better understanding of what it takes to be prepared can go a long way toward preserving both business activity and peace of mind.

“Being unprepared in terms of power and fuel can have drastic consequences for small to midsize enterprises,” says Ashish Moondra, product manager for power management devices at Avocent (www.avocent.com). “In the event of a blackout or even a significant brownout, unprepared businesses may encounter hardware losses, as well as loss of data. They may also have to deal with corrupted operating systems and applications. Time required to recover from such losses may put the unprepared businesses out of action for durations that far exceed even the duration of the power event.”

Don’t Be Caught With Dead Batteries

A critical piece of the puzzle of protecting power assets is the UPS, which can keep sensitive electronic equipment such as servers running for a temporary amount of time when a blackout occurs. According to Matt Kightlinger, director of solutions marketing for the Liebert Products business of Emerson Network Power (www.emerson.com), the exact size and design of the UPS and corresponding cooling equipment should be based on the needs of the particular data center.

The first lines of defense in the event of a blackout, explains Moondra, are the batteries in UPS solutions, but they are the most vulnerable and failure-prone part of these devices. “At the very least, enterprises should ensure that their UPS battery solution includes a parallel string of batteries. In addition to that, it is very important that these enterprises have a preventative maintenance plan for batteries. Such plans are offered by most UPS vendors, as well as by several data center power specialists,” Moondra says.

Stay Adequately Fueled

According to Eric Linxweiler, senior vice president of solution services for Logicalis (www.us.logicalis.com), most midsized data centers are unprepared for catastrophic failures that can lead to a long-term (eight hours or longer) power interruption. Although these data centers might have a generator backup solution, they often rely on outdated requirements and ultimately end up running an underpowered generator that has insufficient fuel for the duration of the event.

“With respect to fuel, ensure you have a contract to deliver fuel for the required time you may be operating on your generator power,” Linxweiler says. “Also, ensure that you have the ability to get fuel from more than one source if you deem the risk relevant to your location. We’ve seen situations where fuel deliveries have been interrupted for up to a week following a regional disaster. This, coupled with commercial power interruption, could bring data processing to a halt.”

Know Your Location

Some data centers, particularly those that are new to a certain region, can be caught off-guard when a power outage occurs. Although the outage itself might not be a surprise, access to the supplies necessary to survive the outage can be surprisingly limited (or the cost can be high) depending on the location of the data center. Such factors should be heavily investigated before an outage actually occurs.

“Oftentimes the access to fuel is a much bigger issue than one thinks. . . . One factor that is often overlooked is the actual cost of power and fuel. From that perspective, California is much more expensive than Washington or Ohio, for example. That’s a larger cost over the lifetime of a data center, which could far surpass the contingency planning costs of generators and fuel,” Linxweiler says.

He also suggests that data centers work with local power providers to determine the scope of their response during power outages. Ideally, two power sources from two different providers should be available for your facility, which can help to reduce single points of failure.

More Are Better Than One

Data centers are no strangers to the concept of redundancy for critical applications, but the redundancy concept also applies to critical power equipment, says Liebert’s Kightlinger. Redundant power systems can be implemented to ensure higher reliability by eliminating single points of failure, he says. And just as servers and other systems can be kept running smoothly with the help of monitoring, so can power systems.

“Real-time monitoring of the data center’s critical infrastructure is recommended,” Kightlinger says. “This functionality allows real-time event notification, such as a power outages, to IT staff so they can assess the situation properly at all times. Remote monitoring often is critical for small companies, since they may not have onsite staff around the clock.”

by Christian Perry