On Sept. 29, the EPA released its updated preliminary data collection results for the Energy Star for Data Centers program. It has decided to call this new metric EUE, for Energy Usage Effectiveness.While at first this would seem similar to The Green Grid's PUE, it is very different in that it refers to the 'energy' contained in fuels, not "power' to the data center, as normally measured as electrical power in kilowatts. Specifically, it is defined as 'total energy/UPS energy.' Moreover, it seeks to factor in 'source energy' as the basis of EUE calculations, which is defined by the EPA as the raw energy in the fuel, e.g. natural gas, coal, hydro, wind, solar (and to differentiate from the actual data center's "site energy," not power).
It is still unclear what this will mean in practice, since no real details were discussed about how this will be applied. Just as importantly, a Data Center's EUE rating will be scored on a comparative percentile basis of 1 to 100 against the group's collective results, wherein 100 is the top score. All applicants, both present and future, will be compared to this first 'reference' group for years to come. So the race is on to get the Energy Star for Data Center Plaque (yes, a real plaque). The score must be in the 75th percentile or higher to pass and become an Energy Star-certified data center.
Of the 121 data centers that volunteered, across 24 states, only 108 contestants qualified to be in the final sample reference database. I say contestants because there will be a prize, an actual Energy Star Plaque, awarded to the top 26 sites. This original survey database of 108 sites will form the basis of comparison for future Energy Star program applicants.
The actual results seemed surprising to all concerned. It was like watching multiple episodes of 'MythBusters' for data centers. Here are some of the top 'myths' revealed by the results that seemed the most surprising:
What I found lacking in all the potential complexity of measuring and calculating 'source energy' is any mention of the significant use of water by data centers, since water is an important resource being 'consumed." (Yes, I know that the water only evaporates and is not really used, but it does requires energy to process and pump it.)
The EPA has been watching the data center market for several years and released its first 'Report to Congress on Server and Data Center Energy Efficiency' in August 2007. It also released the 'Energy Star for Servers v 1.0' in May 2009, which was a very well conceived and executed joint effort with the major server vendors, which should have an important and positive impact on the energy efficiency of new servers.
I am all in favor of thinking of the overall 'big picture' regarding energy use and efficiency, and a lot of good work has been done by many dedicated people on this project. However, there are not many coal-burning data centers, and while a few do have on-site co-generation, the majority are based on conventional local electric utility power. I hope that in the EPA's efforts to come up with a better metric, such as by specifying 'source energy,' rather than The Green Grid's 'simple' PUE, it does not become counterproductive.
In a worst-case scenario, the EUE calculations could become so arcane and complex that potential future applicants would not even try to meet the requirements. In the future, perhaps large data centers might even use nuclear energy or even server rooms powered by a 'Mr. Fusion' (so does anyone out there in IT land know how many BTUs are in a uranium fuel pellet? I mean besides the folks at Lawrence Livermore Labs?).
So in the meantime stay tuned, the EPA is in the final stages of deliberations and anticipates that the final EUE parameters will be released in November.