Tuesday, December 12, 2006

Publishing server energy efficiency ratings

A few weeks back we asked the question: Could an organization publish side by side data on server energy usage based on the EPA's recently annoucned metric for one-unit and two-unit servers? And if so, who should do it?

We polled experts and participants -- here is a sampling of the responses:

Charles King, principal analyst, Hayward, Calif.-based Pund-IT: Up until now, you could go through and reverse engineer [to find the server efficiency], but you would have to be really handy to have some kind of rating. Personally, I think it would make sense for the EPA to put something out like this. You want something where the numbers aren't bent by any vendors' product strategies. I'd be interested to see the overview work that LBNL is doing as well. From everything I've read so far, server efficiency plays a critical role, but it's a small role. So I imagine the work that LBNL is doing will put it in context beyond whose server model draws less power this quarter versus the next quarter.

Gordon Haff, analyst with Nashua, N.H.-based Illuminata: EPA/Energy Star would probably have the most credibility and independence. Among industry organizations, SPEC benchmarks are also quite well accepted. A standard measurement suite and methodology that was harmonized between those two organizations would, I believe, have a high probability of being accepted as a credible standard measurement.

Ken Brill, executive director, Uptime Institute, from his recent Q&A with SearchDataCenter.com: [The Uptime Institute] would be interested in [publishing the data], but we haven't figured out the economic model. There are three basic models. There is the 'Consumer Reports' subscription model. They buy products at the consumer level so you're not getting a hyped-up machine, and the consumer of the information pays for it. Because they pay for it, they feel a certain amount of integrity for the testing that's been done. The second model is the manufacturers can run the testing on their own and publish the results, but then it's not all in the same place, and there's always the suspicion of whether the tests are rigged. In the middle would be tests the manufacturers paid for but which were run by a third party and available through the third party. I think the best way to do that is if it's user-paid-for. As long as it's vendor-paid-for, then it's always subject to some concern. We have the technical resources to do it, but is it something that's going to be financially viable? Personally, I think the 'Consumer Reports' approach is going to be the one that people are going to feel the most comfortable with, but how much will people pay for these reports?

Mark Bramfitt, supervisor of the customer energy efficiency program for the high tech market at PG&E, from a recent Q&A with SearchDataCenter.com: The industry right now publishes servers' top speeds and they test each others' equipment. If you say top speed is 250 miles per hour, and it's really only 200, you can bet that one of your competitors will point that out. I think the industry has a good model for publishing data. If they add the miles per gallon figure, which is what SPEC and the EPA are working on, I don't know why that wouldn't fit in with what they're recording already. Whether that information is all in one place or not, or whether you have to go to the vendors' individual sites, it doesn't really matter. As long as it's reported in a transparent fashion and it's self-policing. It may be that the EPA runs a Web site that runs it all in one place and that would be magnificent. That way a customer doesn't have to pull up 3-4 different Web sites.

Jonathan Koomey, Ph.D., Professor/Staff Scientist, Lawrence Berkeley National Laboratory and Stanford University: My feeling is that the data should be publicly available on the Web, so that customers can easily compare. It's probably better if there is one place that compiles the data. That could be SPEC, it could be LBNL, it could be the Green Grid. I'll be talking with various companies about this in coming months, and will probably have some insight into it early next year. It probably makes more sense for someone who really knows servers to pull together the information, with feedback from AFCOM and other user groups.

Christian Belady, distinguished technologist, HP: HP expects to do its own measurement following the protocols published by established benchmark organizations which reflects our current approach with existing performance benchmarks. HP then would make these results openly available to customers as would the organizations. In addition, HP has also published data center level efficiency metrics that it is currently socializing in the industry.

What do you think? Drop me a comment.


At 6:39 PM, Anonymous Anonymous said...


At 6:40 PM, Anonymous Anonymous said...



Post a Comment

Links to this post:

Create a Link

<< Home