Tech giants push the green data center
In catching up on Green IT news the other day, I came across an announcement about a data center of the future project from T-systems and Intel. T-Systems, a division of Deutsche Telekom, and Intel, have dubbed the project DataCenter 2020.
The partners aim to improve data center energy and cooling efficiency by measuring variables in this test setting. The data center has sophisticated gear such as sensors to measure humidity, a smoke generator to visualize air flows, and an adjustable ceiling height. Obviously, not many companies can afford or justify such advanced gear in even a large data center, but surely there will be lessons learned from this test setting that could be applied to data center products and configurations.
New to me from reading about DataCenter 2020 is something called PUE, which stands for power usage effectiveness. PUE, originated by a group called the Green Grid, is the ratio of the total energy consumed in a data center to pure computer operation. The current PUE values for data centers cooled with recirculated air range between 1.7 and 1.8 on average, according to T-Systems, but the DataCenter 2020 project aims to deliver a PUE value of 1.3.
With PUE, lower is better. According to this blog post on DataCenter Knowledge, an average PUE of 2.0 indicates that the IT equipment uses about 50 percent of the power to the building. A number of data center operators, including Google, have announced PUE readings of 1.2 or below for their data centers.
Data center experts (see blog mentioned above) seem concerned about applying PUE in a more consistent, practical way in industry, and not over hyping one organization’s PUE numbers in way that might be unfair to others who measure PUE in a more rigorous way. This seems rational. In fact, the PUE debate reminds me of the gnashing of teeth people in manufacturing operations have had over useful application of the overall equipment effectiveness (OEE) metric. Similarly, OEE can be applied inconsistently between organizations, or even within the same company. That doesn’t mean it’s a bad metric, especially if those who use it are consistent it its application, and don’t lose sight of other objectives and metrics. At any rate, it’s encouraging to see debate aimed at making PUE a more useful metric, and to see big tech companies like Deutsche Telekom and Intel investing in a test bed to bring data centers to the next level of efficiency.
Roberto Michel commented:
Thanks Larry, for pointing out The Green Grid has considered issues around consistent use of PUE, and has provided guidance. I was unaware of that. To others, please check out TGG's site!
Larry Vertal commented:
Roberto,
I enjoyed your post on the recent DataCenter 2020 project. In case you weren’t aware, Intel is a founding member of The Green Grid. We have been gratified by the use of PUE globally since it was introduced by The Green Grid (TGG). And Intel like many of our nearly 200 member companies along with datacenters worldwide have been using PUE extensively in efforts to advance data center energy efficiency. As you infer metrics often introduced for the purposes of continuous improvement of a datacenter often get grabbed to try and do comparisons between one datacenter and another. Given this reality The Green Grid, since the introduction of PUE, has developed further guidelines for reporting and also has released to the public The Green Grid website a reporting tool that allows datacenter operators to report their PUE in alignment with these guidelines. We hope The Green Grid’s PUE and DCiE reporting guidelines will lead to an even greater industry adoption and we’re always looking for industry involvement and encourage organizations to join and contribute their ideas to The Green Grid.
Thanks!
Larry Vertal, Executive Director of The Green Grid
www.thegreengrid.org
