Data Center Efficiency: Is It Fair to Compare?

PUE (Power Use Effectiveness) is meant to measure efficiencies of a single facility so owners can determine where energy is being lost.

There’s a discussion going about PUE metrics and whether or not it is fair to compare efficiency ratings of different facilities.

Industry insiders suggest that PUE is not meant to compare one facility to another. If in fact, the PUE of similar data centers are to be compared the following factors must be examined:

Data Center Tiers

Data center Tiers represent the reliability of a facility. A Tier 1 data center will have different energy capacities than say a Tier 4 facility. Tier 4 facility is the highest enterprise class facility that likely has greater network storage demands. Most likely, a facility like this will have a higher PUE rating based on the power needs of the equipment. Such a facility may have a PUE rating of 3 based on redundancy but it is still something that is continually monitored to see how much energy is lost in relation to how much of it is consumed.

Other Factors Affecting PUE

Regardless of the size of the facility, climate, age and business model play a significant role in determining a facility’s PUE rating. Colocation facilities typically require a lot of power, redundancy and cooling capacities. So, it is unlikely that a colo facility will have an more efficient PUE than a modular data center – as load is more easily scaled to meet demand. A data center located in a cold climate will require less energy to cool itself just as an older facility will likely consume more electricity as its aging hardware struggles to keep up with computing demands.

Defining Efficiency Standards

PUE (Power Use Effectiveness) is meant to measure efficiencies of a single facility so owners can determine where energy is being lost. The lines between industry standard and benchmarks for each facility have been skewed, which is why this debate is ongoing. Industry folks stress that PUE isn’t meant to compare efficiency standards across the broad range of data center facilities. It should simply be used as a way to improve energy efficiency within a specific condition.