The boss called, “What’s our PUE?”
Do you know your PUE?
PUE stands for Power Usage Effectiveness and it is rapidly becoming the number to know. In the past, datacenter managers were simply asked to provide enough space, power and cooling to support the IT equipment. Now, the same managers are being asked to do it efficiently. PUE can be a helpful benchmark.
Introduced by the Green Grid, PUE is a measure of efficiency. It is defined as: the total facility power consumed divided by the total IT equipment power consumed. The total facility power is measured at the utility meter for datacenters. (For mixed-use facilities like an office building that contains a datacenter, only the power needed for the server/datacenter room should be measured or even estimated.) The facility power includes everything that supports the IT equipment including power, cooling, lighting, etc.
The IT equipment power is the load associated with servers, storage, networking, workstations, etc., that are used in the datacenter. Of course, the total facility power will always be greater than the power required by the IT equipment. So the PUE calculation will always be greater than one. But, how much greater?
A PUE of 1.0 would be an ideal situation: no power distribution losses, no chillers, pumps or fans, etc. While that is not possible, industry giants like Microsoft and Google are planning for PUE’s of 1.2 or even better. That would truly be best in class. Today, according to the Uptime Institute, a typical data center has an average PUE of 2.5. However, it is not uncommon to have a PUE of 3.0 or greater. This means the IT equipment consumes only 1/3 of the power. Or, put another way, 2/3 of the power (and the utility bill) is wasted!
Whether you are looking to go Green or lower your OpEx, knowing your PUE is the first step. From there a step-by-step plan can be put in place to move that number down. So, measure now – before the boss calls.
Do write to me if you need any help on how to go about measuring your datacenter’s PUE.
- » Kubernetes and multi-cloud: How to monitor your modern applications effectively
- » Is performance engineering still needed when it comes to cloud?
- » Sainsbury’s looks to Google Cloud for machine learning as retail cloud case studies continue to climb
- » Amazon completes consumer database migration from Oracle to AWS
- » AWS, Azure or Google: Do the differences between cloud providers really matter?