Power Usage Effectiveness (PUE) has become a recognised standard metric to help determine the efficiency of a data centre’s infrastructure, and a vital tool for implementing an effective energy management plan. However, there are many issues that cause difficulty when calculating a figure for individual data centres. Understanding how to account for the various power-drawing elements is crucial to producing meaningful figures to benchmark data-centre efficiency.

Problems in calculating PUE include whether and how to account for the power drawn by certain devices in a data centre; how to account for subsystems that may be present in some centres but not in others; how to calculate figures for elements in shared facilities whose power output is not entirely “charged” to the data centre; and how to account for the power consumed by subsystems that are either impractical or costly to instrument for power consumption.

“Users can always make up their own approach to collecting and processing energy data, but data centre efficiency cannot be benchmarked without a standard method,” said Matthew Baynes, enterprise sales director, Schneider Electric.

Authored by Victor Avelar, director, senior research analyst, Data Center Science Center, Schneider Electric, the White Paper defines a standard approach to collecting data from data centres and showing how to use it to calculate PUE, with a focus on what to do with data that is confusing or incomplete.

White Paper #158 describes a three-part methodology to standardise the calculation of PUE. It establishes a standard to categorise data centre subsystems as IT loads, physical infrastructure loads or loads that should not be included in the calculation. It describes a standard methodology for estimating power consumed by subsystems shared with non-data-centre loads. It also describes a standard methodology for estimating the power consumed by subsystems where individual metering is impractical.

The White Paper makes the point that various standards bodies such as the Green Grid and others recognise the problems outlined in the paper and are working on solutions in the form of guidelines and standards. As such, White Paper #158 will be updated as and when new standards emerge.