Data centre monitoring is essential when calculating the accuracy of Power Usage Effectiveness (PUE) — an industry-accepted energy efficiency metric for data centres — according to a new White Paper from energy management and automation specialist Schneider Electric.
Depending on business requirement, location and cost there are a number of options available for acquiring the necessary data to calculate an appropriate figure for data centre PUE, which requires accurate measurements to be taken of the power consumed by IT equipment and separately from the overall power consumption of the entire facility. And the appropriate method should be determined purely by the business objectives for measurement.
Schneider Electric's new White Paper says the most accurate method is to take continuous measurements of the power expended by the various IT and infrastructural elements in a data centre to calculate how PUE varies over time in accordance with changing IT load conditions, available support infrastructure systems and especially the outside climatic conditions. Even so, there are various options available for continuous metering and monitoring, which will depend on the particular purpose and surroundings of each data centre.
The various factors that should be considered when choosing how to acquire, aggregate and analyse data on a continuous basis for accurate calculation of PUE are discussed in 'Continuous Metering and Monitoring of PUE in Data Centers' , which considers the drivers that can affect how an organisation chooses to deploy the most appropriate data acquisition and aggregation tools for its needs.
Four levels of reporting
It also discusses the four levels of PUE reporting as set down by the Green Grid — Level 0 does not cover continuous monitoring at all; Level 1 requires only monthly measurements and is intended for situations in which only the IT equipment in a data centre is powered via the Uninterruptible Power Supply (UPS); Level 2 requires daily measurements and takes a more granular approach to measuring power consumed by the IT equipment by measuring at the PDU output.
This is important in cases where the UPS also supports infrastructure equipment such as air handlers; Level 3 reports IT load most accurately, as the meters are at the IT rack level. It also comes with added complexity, however, as there are many more meters that must be aggregated together. Level 3 ultimately helps the user understand why the PUE is trending up or down over time, and most important, what can be done about it.
The paper goes into some detail as to how costs can be minimised even as one deploys the greater number of meters needed for the most accurate data acquisition possible. It also discusses how additional metering can be installed in a working data centre without causing downtime and thereby incurring unnecessary costs. And finally it demonstrates the various methods that can be deployed to aggregate all the data from several measurement points using Data Centre Infrastructure Management (DCIM) management software and the trade-offs that should be considered before deciding on the most appropriate course.