Green Grid Seeks to Rein In Misuse of PUE Energy Efficiency Rating

Green Grid Seeks to Rein In Misuse of PUE Energy Efficiency Rating

Power Usage Effectiveness, or PUE, has become something of an untamed beast in the data center industry. The standard, developed by the IT industry group The Green Grid and released to the public four years ago, has become the de facto standard of energy efficiency -- a "miles per gallon"-type rating for green data centers.

Now the standard has fallen victim to its own success.

Data center owners and operators are constantly touting their PUEs -- a number that measures how much of the energy entering data centers is providing compute power as opposed to lighting, cooling and other support services -- in what on the surface is a promising "race to the top" for the greening of data centers.

The problem lies not with the standard, but with how it is applied. Not only did the original PUE guidance published by The Green Grid in 2007 leave out detailed methods for how to measure PUE, it also neglected to clarify that PUEs are not a reliable basis for comparing two data centers.

"PUE is the most-used, most abused metric out there," Vince Renaud, Vice President of The Uptime Institute's Professional Services, explained last week at the Uptime Institute Symposium. "Use it for what it's supposed to be: individual data centers, measuring against yourself, not against this one over here or this one over there. It's gotten completely out of control."

A task force consisting of members of The Green Grid, The Uptime Institute, Energy Star, the U.S. Green Buildings Council, and other groups, yesterday took steps to rectify the problems with PUE, publishing a new white paper that clearly delineates how to measure PUE, giving the IT industry four different ways to measure energy efficiency, with varying degrees of accuracy.

The white paper introduces a couple of useful additions to the standard. First and foremost it spells out exactly how PUE should be measured, but breaks the standard down to four categories, ranging from relatively simple measurements that provide a performance snapshot to more accurate and sophisticated measures to provide detailed performance data.

The new PUEs, which will be designated PUE0 through PUE3, start with a simple measurement of peak IT energy demand divided by peak total energy demand (PUE0), and advances through to measuring energy at the input to servers, storage or network devices divided by total annual energy (PUE3)

The white paper also gently but clearly spells out that PUE is not a comparison tool:

It should be noted that caution must be exercised when an organization wishes to use PUE to compare different data centers, as it is necessary to first conduct appropriate data analyses to ensure that other factors such as levels of reliability and climate are not impacting the PUE.

In other words, a data center located in Phoenix that houses non-mission-critical computing functions should not be compared to one in Toronto that has multiple levels of redundancy and uptime.

Though the white paper serves to clarify PUE measurement, It's likely that it won't simply end the misuse of PUE ratings. But in solidifying the standard and urging further improvement on data center energy efficiency, The Green Grid has continued its efforts to keep the race to the top on the right path.