Energy and environmental performance of data centers: striving for excellence

A first approach is to carry out an overall assessment of a data center’s energy and environmental performance along three dimensions: the technical environment, IT equipment, and the hosted IT services.

When it comes to technical equipment, PUE (Power Usage Effectiveness) is unquestionably the most widely used to date: it is the ratio between the energy consumed by IT and the total energy used by the data center or, in other words, a ratio used to measure the energy efficiency of the infrastructure supporting IT production. Another example is CUE (Carbon Usage Effectiveness: the ratio between the CO2 equivalent of the facility’s energy consumption and the total energy consumption of IT). Within this scope, these KPIs remain limited because they only take energy consumption into account. To take a global approach, it is necessary to consider potential recovery of the waste heat produced, as well as the share of renewable energy used (REF).

Lastly, energy and environmental performance indicators can be defined for the digital services delivered by the data center’s IT equipment. This involves defining indicators per functional unit specific to the service delivered, such as a virtual machine (VM), for example.

However, measurements within the data center do not make it possible to assess the energy and environmental impact of the IT services hosted there. In addition to overall measurements and indicators for data centers, it is therefore recommended to precisely measure the impact of each service provided within the information system, taking into account the entire life cycle of the equipment supporting that service. This implies calculating, for each infrastructure or IT asset (cooling system, main LV switchboard, UPS, generator sets, servers, etc.), its impact from raw material extraction through to end-of-life recovery, including manufacturing, distribution, and use.

For information, at the scale of a 600 m² data center consuming more than 11 GWh/year, IT equipment accounts for 75% of the data center’s environmental footprint over the entire life cycle.

This life-cycle-based approach to assessing environmental impacts provides important decision support for IT departments. It can, for example, lead to the decision to virtualize an application that is not yet virtualized in order to reduce its energy consumption, to redefine the equipment renewal strategy, or to remove certain applications that are no longer used but still consume resources.

Whatever approach is chosen, given the multiplicity of indicators, it is up to each company to determine—based on its context—what needs to be monitored and measured, as well as the monitoring, measurement, analysis, and assessment methods, and the frequency of monitoring and measurement.

The challenge in measuring energy and environmental performance is to fine-tune metrology using appropriate measuring instruments as close as possible to the equipment, with the information then used as input data for calculating the defined KPIs. The aim is to promote automatic data collection to facilitate analysis of results, rather than relying on manual readings, which are less reliable and time-consuming for teams.

The metrology implemented in data centers is a management tool, not an end in itself: it involves defining target objectives, monitoring whether they are achieved, or conversely identifying potential deviations, as part of a continuous improvement approach.

A final approach to assessing a data center’s energy and environmental performance is holistic, as part of a continuous improvement process. The objective, in this case, is to "benchmark against yourself" over time, in order to strive for energy and environmental excellence. To achieve this, organizations have standards at their disposal such as ISO 14001 (environmental management system for organizations) and ISO 50001 (energy management), ISO 30134 (which defines key performance indicators for information technologies and data centers), as well as EN 50600 (Information technology – data centre facilities and infrastructures). Lastly, the less demanding but still beneficial European Commission Code of Conduct for Data Centres is one of the frameworks endorsed by many companies.

Whether global or more granular, all of these measurements of energy and environmental performance are based on a more or less voluntary continuous improvement approach within each organization and each data center.

However, the 2021 Finance Act changes the energy taxation of data centers as of January 1, 2022. From now on, eligibility for the reduction in the Domestic Tax on Final Electricity Consumption will be conditional on implementing an energy management system and joining an official best-practice framework in the field of energy. Data center operators should therefore, in the coming years, become increasingly likely to commit to these holistic approaches to measuring and continuously improving energy and environmental performance.

By Soumia Hattali, sustainable IT consultant, APL.