![](/MC/Home/Images/epa1sm.jpg)
![](/MC/Home/Images/epa4sm.jpg)
More than 130 people dialed into a 90-minute webcast on May 7th during which the U.S. Environmental Agency (EPA) outlined its progress towards establishing an Energy Star rating for data centers and also revealed some preliminary findings.
Technical difficulties marred the webcast and prevented CSE’s Bill Angle and Alexandra Sullivan of the EPA from fielding questions and opening the preliminary findings for discussion and explanation.
The EPA previously settled on PUE as the metric on which the rating system would be based. The agency has gathered data from 115 data centers but is able to use data from only 90 facilities so far. Of these 90 centers, 34 facilities could use either UPS output or PDU input values to represent IT power. The EPA questioned why the PDU input would yield higher PDUs than the UPS value.
![](/MC/Home/Images/epa5sm.jpg)
The EPA anticipated three results from this debate:
- Allow the use of the PDU input value for Energy Star purposes
- 2. Develop a constant to normalize the values
- Require the use of UPS values for Energy Star purposes.
The EPA said that it wants to strike a balance between creating a useful standard and requiring facilities to install meters on their UPS. Approximately 10 percent of respondents could meter PDU input but not UPS output.
The EPA also noted that lower PUEs seemed to be associated with
- Higher total UPS energy
- Higher UPS utilization
- Lower Tier values (except Tier 4 betters Tier 3)
- Hosting facilities
Findings such as these lead the EPA to ask itself whether the sample set overrepresents traditional data centers as well as how to account for the effect of UPS utilization and data center size.
The EPA announced that it would be
- accepting data through June 15th, 2009,
- performing regression analysis during the summer
- holding a fall 2009 meeting to discuss findings
- unveiling the model in January 2010.