JISC

Overview

While measuring the energy consumption in the data centre is generally accepted as “a good thing”, questions remain about the detail of where to measure, and the degrees of accuracy and precision appropriate for the purposes required. This project will carry out experimental analysis by conducting measurements with varying levels of detail. The outcome will be information and advice on appropriate data centre measurement activity, which will be both efficient and effective.

Aims and objectives

To undertake this project, we aim to monitor the power consumption in different parts of our data centre, varying the metering methodology during the project, so that some metering will be done at high granularity (measuring individual devices and locations); whereas in other cases, the metering will be done at lower granularity (giving the combined energy use of a number of devices and locations from a single meter).

Specific objective include:

• Establishing baseline energy measurement methods;
• Evaluating the effects of different granularities of metering on the baseline;
• Investigate the effect of consumption reduction methods.

Project methodology

The project team comprises a mixture of academic, technical and estates management colleagues. We will draw on academic skills to design the measurement experiments and interpret results; technical knowledge to deploy and maintain the test equipment, while our estates colleagues will advise on and deliver the measurement and data capture facilities.

Anticipated outputs and outcomes

Through research we will identify at least two candidate date centre energy consumption improvements and provide detailed and accurate data on the effect they have on power usage, as measured at varying levels of granularity.

Technology / Standards used

One of our initial project objectives is to review and determine appropriate forms of technology to be used in our project. “Appropriate” in this context means that the technology needs to be representative of the types of system likely to be used; that the power measurement techniques are accurate and reliable, and that the overall experiment is repeatable. We will also explore the range of possible measurement and data reporting techniques in order to contribute to the development of standard forms of measurement and reporting.