One of the greatest strengths of the PUE metric, the industry standard for measuring data center energy efficiency, is its simplicity: Calculate how much energy your data center is consuming overall, then divide that number by how much energy your IT equipment alone consumes. The resulting number is your PUE, which you can then compare against your previous scores or those of rival data centers. The lower the number, the more efficient your data center and the greater the bragging rights.
At the same time, the simplicity has its shortcomings. For example, it gives operators much flexibility as to where to measure consumption -- at the PDU or at the point of connection of IT devices -- as well as how often to take measurements. Thus, two similarly sized data centers might boast nearly identical PUEs, even though one has a highly efficient DC-based power distribution system and measures PUE continually, while the other has a traditional AC-based system and measures PUE once per month during off-peak hours.
[ Learn more about the ROI of green IT -- and discover the sustainability secrets of InfoWorld's 2010 Green 15 winners. | Keep up on green IT trends with InfoWorld's Sustainable IT blog. ]
In an effort to overcome this drawback, The Green Grid has unveiled four categories of PUE, ranging from Category 0 to Category 3, in a new white paper, "Recommendations for Measuring and Reporting Overall Data Center Efficiency" [PDF]. With each level, the measurements become more granular and the results more precise. Thus, a data center operator may choose to go with Category 0, which requires the least effort and fewest resources -- but then those results won't be viewed in the same light as a rival's Category 3 PUE figure.
Specifically, measuring your Category 0 PUE would entail figuring out your data center's total peak load at the utility meter over a 12-month period, typically reported as demand kW on a utility bill. You would then divide that figure by the demand reading of your UPS system's (or systems') output as measured during peak IT utilization.
