Cookie policy: This site uses cookies (small files stored on your computer) to simplify and improve your experience of this website. Cookies are small text files stored on the device you are using to access this website. For more information on how we use and manage cookies please take a look at our privacy and cookie policies. Some parts of the site may not work properly if you choose not to accept cookies.


PUE is still relevant – it just needs a consistent methodology

Since the Green Grid first suggested it, PUE (power usage effectiveness) has taken its place as the standard metric for data centre efficiency. PUE was developed to help make data centres more efficient, and a key industry benefit has been to bring the principles of energy efficiency to the discussion table.

The data centre industry understands that PUE is not perfect. However, it does provide a simple and practical matrix for measuring relative efficiency of power use inside the data centre. It allows data centre operators to generate a matrix of key criteria, recorded along a timeline to provide comparison data. So, a relative increase or decrease in efficiency can be tracked and remedial action taken accordingly.

Questions of science

Vintage electric meter

Vintage electricity meter

Source: Thinkstock / ac72

Part of the problem that ASHRAE may have is the unscientific nature of the current PUE metric and how it is used depending on where in the world the data centre is sited.

There are also questions about which aspects of the physical infrastructure should be measured and which left out of the equation. We are, I’m sure, all aware of how PUE was quickly adopted by marketeers to evidence facility efficiency. All that was needed was the purposeful elimination of certain plant’s energy use to generate an impressively low PUE

On the plus side, used correctly PUE can be a good indicator of where you are in the management of a facility. It certainly can be an effective tool if the same methodology is applied to measure the same data centres or the same group of data centres. You can effectively see how you can improve and rate the level of improvement. Using the same criteria provides a repeatable methodology and therefore results that are comparable.

However, there are circumstances where PUE becomes a difficult metric to align to the rating of a specific data centre. For example, a Tier IV data centre. It is also not exactly perfect for making comparisons between facilities due to the different ways in which the metric is applied and a vast number of variables affecting the result – like ambient weather conditions which can be different from location to location.

Using the same PUE methodology in measurement gives you an indication of how efficient your facility is. As an operator, it helps you to understand how efficient your processes are and how you can improve the efficiency. However, if you are comparing let us say, ‘colo-sites’ then you have to take a wider range of variables into account, not least what your goals are and the relative importance of each of the variable elements in your decision making process.

Measurement tools make sense and PUE fundamentally is useful as long as the criteria is understood and you know what you want to achieve with it. You cannot simply take two sites’ PUE and make operational decisions based on those figures.

In the process of constantly striving for more efficient data centres there are new technologies that provide real-time data flows to the management of the data centre. Systems such as SynapSense offer a dynamic approach to real-time understanding of the energy efficiency within the data centre. This provides not only the data to continuously improve energy efficiency, but also automated control of cooling systems to effectively control energy usage and minimise PUE figures.

Michael Adams is a director of EMEA IDC business at Panduit.

Readers' comments (1)

  • PUE does not account for energy overhead within the server box which can add up to 30% of the motherboard consumption at maximum power.

    Even though modified power trains and liquid cooling reduce the in box overhead to less than 5% the PUE could little different from a fee air cooled DC despite the fact that the liquid cooled data center was drawing less overall power.

    If you want to make PUE work, server suppliers would have to state the PUE of thier servers, that is, the power and cooling overhead within the server box itself. The figure could be the ratio of the server power input to the motherboard input minus the fans (if they are driven through the motherboard).

    The real PUE would then be facility PUE times server PUE.

    Unsuitable or offensive? Report this comment

Have your say

Please view our terms and conditions before submitting your comment.



More link