ENERGY STAR Data Center Data Collection Preliminary Result |
Posted By Zen Kishimoto,
Monday, October 05, 2009
|
In May, EPA released the first EnergyStar for server specification and about the same time started work on EnergyStar for storage. Along with these efforts, EPA has been collecting data from various data centers in an attempt to create the EnergyStar rating for data centers.
The preliminary results were presented on September 28 webnair by EPA. (The presentation was distributed to the webnair participants and will be available in their web page, Enterprise Server and Data Center Energy Efficiency Initiatives. )
Initially, I did not plan to report this on my blog because I thought other people would report it. But to date, I have not seen much of the coverage except for two:
From the datacenerdynamics article:
EPA will use a data center’s EUE (Energy Usage Effectiveness), normalized for a variety of the facility’s characteristics, to calculate the 1-100 rating for that facility. Data centers in the top 25 percentile (ones that receive a 75 or higher rating) will qualify for Energy Star certification….
EUE is calculated by dividing total source energy by total UPS energy. The EPA is using source energy instead of site energy to be able to account for a variety of fuel types a facility uses in its energy efficiency equation. Those fuel types may include things like chilled water delivered from the utility. Source energy also accounts for losses that result from conversion and transport of energy to the site.
One thing noteworthy here is that EPA distinguished energy from power and used source energy rather than site energy. This means all the energy required from source to the site, including any loss (including transmission loss) and anything required to generate that energy. As far as I know, this is my first time to hear a term, EUE, as also reported by other reported and bloggers.
The datacenerdynamics article summarizes the important points but there were more details and I would like to touch on them.
The timeline of this effort is as follows:
- October 2007 to March 2008: Preparation
- March 2008 to June 2009: Actual data collection
- June 2009 to November 2009: Analysis and adjustments
- Another Webnair to discusses the next version based on the solicited comments on the current version: Early November
- April 2010: Publish the Energy Star for data center specification
So far, 121 data centers submitted (from 24 states) their data for analysis, consisting of various types (size and tiers).
Due to the overwhelming majority (power measuring location: UPS vs. PDU), the energy consumption observed at UPS equipment is used for IT energy consumption. The average EUE for the submissions were 1.91 and the range of EUE among the submitted data was between 1.25 and 3.75. EPA tried to correlate the specification more with energy management and less operation practices.
Data center analysis model is being developed and four factors may be included are (because they appear to be relevant to the analysis):
- UPS energy intensity
- UPS energy (IT total energy)
- Square feet of a data center
- Tier
But the following factors will probably not be included:
- Heating Degree Days (HDD)
- Cooling Degree Days (CDD)
- Data Center Type (traditional, hosting, internet, etc.)
- UPS Utilization
Expanding the above four factors that may be excluded:
- Climate
- No observed effect of climate on energy consumption
- Believed to be due to the fact that Internal loads are much higher than Climate loads
Type of data centers
- Does not seem to make a lot of differences
UPS utilization
- EUE is slightly lower in buildings with high UPS Utilization, as expected
- Only significant in regressions if UPS Energy is excluded, but not nearly as strong as UPS Energy
One other interesting point made during the presentation was no impact by economizer in the cold climate. A participant from HP was not convinced with this finding as he had a project that showed an opposite result.
In summary,
- With over 100 data centers, EPA has identified that there will be adequate data to support a 1-to-100 Rating
- The EPA rating will incorporate energy usage effectiveness (EUE)
- Defined as Total Energy / UPS Energy
- The data show a strong correlation between EUE and total UPS energy
- This effect will likely be included in the rating algorithm
- The data show a weaker correlation between EUE and Tier Level than anticipated
- This effect is still being evaluated for inclusion in the rating algorithm
- Data centers do not exhibit a strong weather dependence
EPA would like to solicit comments on their current work. Those who want to do so are encouraged to contact them via ENERGYSTARdatacenters@icfi.com.
This post has not been tagged.
Permalink
| Comments (0)
|
Print Page
RSS