When it comes to reliability in the data center, power isn't the problem. As one consultant told me at the 7x24 Exchange Spring Conference, power isn't the issue anymore. Data center operators really just need to point where the power needs to go. If that sounds like hyperbole, consider that Pacific Gas and Electric is dealing with requests for data centers in its service area in excess of 20 megawatts.

Cooling, it turns out, can be an even bigger reliability problem in today's high-density data center than power. Consider one person's story during the conference regarding a simulated an outage. The UPS kept the servers running. But in less than two minutes, the temperature climbed above 104 degrees because the data center wasn't designed to keep the everything properly cooled during a power loss.

In a real world scenario, that means even though the UPS might be designed to carry the power load for 15 minutes, an operator would be forced to shut the entire data center down in less than 2 minutes or risk server failure from surging temperature.

One company said that it uses cold storage tanks and pumps in its high-density data center. The tanks allow operators to keep the temperature under control if there is a power outage, and if necessary, shutdown servers more gracefully.

Until you do a test though, it's hard to know exactly how problematic cooling will be during a power outage. Testing could be especially important if your configuration has change since the initial design. If an outage occurs, will you be feeling the heat?