PI Unskewed

Part 3: Providing a suitable visualization of data center cooling performance, through the PUE ratio (PUEr)

If PUE is the best measure we have for data center efficiency, and if we choose to look at a scale of 0-100% (for the sake of displaying metrics on the Performance Indicator) that brings us to DCIE (Data Center Infrastructure Efficiency).

But all in all, efficiency is just one piece of the PI, and applying DCIE as the PI measure of energy efficiency – for all data centers – would be misleading.

Take, for instance, an enterprise facility that cannot afford to risk downtime, and to defend against it, may be willing to sacrifice efficiency. Yet a PI score with 97% IT Thermal Resilience and 51% DCIE puts the focus on inefficient performance, when it really should be on the 3% loss in resilience!


Clearly, a target range for DCIE extending to 100% is not representative of the business, and the focus is inappropriately skewed.  

PI Energy Efficiency Bands

For PI, much like what we see on washing machines, kettles and buildings, PUE has been categorized into energy efficiency bands, from A (the greenest band) to G (the reddest one).

You can choose any color you like – the band echoes the type of facility, the climate and most importantly, your business requirements.

Green is the color for hyperscale facilities with high hopes for efficient performance, but our example enterprise facility – with its traditional mechanical cooling and PUE of 1.96 – is likely aiming for band D.

They can now readdress their target range – showing the PI score to be more reflective of the business.

PI Realistic Target DCIE

Although the targets are now more realistic, the view of resilience is still eclipsed by the inefficiency.

What if we instead measure where actual PUE falls within the band that the business realistically aspires to operate within? i.e. PUEr(D).
That ratio is the PUEr.

And the result? A PI score that is reflective of business needs, and performance metrics that are no longer skewed.


In conclusion, PUEr simply relates PUE to your business needs through energy efficiency ratings. PI is not changing the way we calculate PUE, just how we present it.

PI Rainbow


Blog written by Maira Bana, Consultant Engineer

For instructions and a detailed example on how to collect the data, generate CFD simulations and calculate the metrics, see the Performance Indicator whitepaper by The Green Grid.

More to come: In the next blog, we look at how we calculate PUEr.

Related Links:
Part 1: PI: Solve it Like Magnum
Part 2: Say P.I. Again

Other Recent Posts

Data Center Simulation: The Best Method To Safeguard Your Data Center

You’ve taken the necessary steps to protect your data center from thermal failure; you’ve built red…

Read More

Beat the Heat with Purpose-Built Data Center CFD

Data center operators in Europe are unsure whether their facilities are properly equipped to handle…

Read More

27 September, 2017

Back to entries