5 best practices for monitoring the environmental footprint of data centers

Like other strategies (commercial, economic, financial, etc.), CSR (Corporate Social Responsibility) initiatives—one of whose challenges is implementing actions to reduce environmental impacts—are only effective if objectives are defined using the SMART approach:

• Specific, meaning they have a clear, tailored objective. The more precise the objective, the more targeted the efforts required to achieve it can be.
• Measurable : measurable data makes it possible to verify whether objectives have been met or not.
• Achievable : the objective must be ambitious in light of the skills available and the resources provided.
• Realistic : an overly ambitious objective disconnected from reality can lead to the strategy failing.
• Time-bound : finally, the objective must have a deadline.

Thus, for example, “reduce the environmental footprint of IT infrastructures by 50% by 2030” is an objective that could be described as SMART. In the context of operating a data center—multifactorial by definition—we propose five best practices to implement for SMART monitoring of the environmental footprint of these IT infrastructures.

The data center load factor is the ratio between the electrical power actually used by IT equipment and the total power available to supply that IT equipment. Good design and proper sizing of IT resources therefore contribute to energy efficiency and reduced infrastructure costs. To achieve this objective, the following additional actions can be implemented:

• Analysis of the activity of deployed physical and virtual machines, such as their number and type, frequency, resources used, etc.
• Taking into account the impact of operational constraints, such as SLAs or contractual service levels.
• Identifying resource issues: saturated network bandwidth, overloaded physical machines, or, conversely, underutilized equipment to be powered down…
• Replacing faulty resources with servers natively designed to optimize power consumption, such as servers from the Open Compute project (The Open Compute Project).

It cannot be said often enough: monitoring and making important decisions requires measuring relevant indicators. Therefore, the first step in an approach to reducing environmental impacts is to quantify those impacts.

Proven and recognized for its effectiveness, life cycle assessment (LCA) is a standardized method used to measure the quantifiable effects of products or services on the environment. LCA takes into account all stages of a system’s life cycle (manufacturing, use, and end of life) in its assessment.

The principles, requirements, and procedures for LCA are defined by the international standards ISO 14040 and ISO 14044. It is a multi-criteria approach based on several indicators for analyzing inputs and outputs—namely everything that goes into manufacturing the product (iron, water, oil, gas resources, etc.) and everything that comes out in terms of pollution (waste, gaseous emissions, discharged liquids, etc.).

Life cycle assessment is carried out taking into account the activity of the data center being analyzed and its challenges. To avoid misinterpretations linked to pollution shifting, the life cycle assessment must include environmental indicators such as:

• energy consumption,
• greenhouse gas emissions,
• water consumption,
• depletion of abiotic resources,
• electronic waste generation.

LCA provides an overall view of the environmental impacts of IT infrastructures and helps identify improvement levers and the system elements that should be prioritized to reduce the environmental footprint.

A good practice is to carry out this type of study every two to three years to ensure that the initial improvement areas have delivered the expected effects after implementation.

Once the life cycle assessment has been carried out, the identified improvement levers must be monitored continuously through simple environmental indicators based on consumption (water, energy, etc.) or more complex ones based on efficiency measurements:

• Electricity consumption

This is a very easy indicator to track because it is reported by the electricity supplier, which bills based on actual consumption. Very often, electricity consumption is also monitored at the level of IT equipment or major uses such as air conditioning, ventilation, or electrical protection.

• The PUE (Power Usage Effectiveness)

PUE is the best-known indicator. It measures the energy efficiency of the IT infrastructure—namely the consumption of the IT hardware itself, but also the energy required to cool the equipment and other ancillary needs.

It is the ratio of the total energy consumed by the equipment (IT equipment and ancillary systems) divided by the energy dedicated to IT production. PUE helps estimate the energy cost associated with the infrastructure. It provides an estimate of the effects of variations in energy consumption. PUE is unitless: the closer it is to 1, the more efficient the data center is considered to be.

• The CUE (Carbon Usage Effectiveness)

This indicator complements PUE. It measures the greenhouse gas emissions induced by the operation of the IT infrastructure.

It is the ratio between total CO₂ emissions linked to the total energy consumption of the data center and the energy consumption of the IT equipment. This indicator helps operators choose more efficient technologies in terms of greenhouse gas emissions. It also makes it possible to assess the relevance of improvement actions to be implemented to reduce greenhouse gas emissions. It is expressed in kg CO₂ per kWh IT.

All these indicators complement one another. They must be measured and interpreted together in order to identify how effective actions are in improving the data center’s overall environmental performance.

Knowing how to measure indicators is good. Ensuring regular monitoring is better—especially when automated in a dedicated dashboard.

The dashboard provides an overall view of all relevant indicators to analyze for monitoring and managing environmental performance, and removes uncertainty. It is also a real decision-support tool: indicators, compared with reference values, make it possible to understand the infrastructure’s environmental health, diagnose the causes of unmet results, analyze gaps, and undertake appropriate mitigating or corrective actions.

Monitoring the environmental performance of IT equipment can also involve using tools such as DCIM (Data center Infrastructure Management), designed to manage the IT infrastructure within a single tool using common processes: change management, incident management, maintenance management, etc. This is in order to extend ITIL best practices across the entire chain of elements, from IT equipment to the data center infrastructure.

Monitoring the environmental footprint of IT infrastructures must also take into account the available human resources: with awareness and proper training, teams can then fully engage in the overall approach to monitoring environmental performance.

This also requires excellent communication between stakeholders, who are numerous and varied: systems and network engineers, developers, etc. Long-term, multi-channel communication: on-site displays, information on best practices, team challenges, internal competitions, hackathons, and awareness campaigns on environmental issues.

Beyond raising awareness and securing the involvement of stakeholders within their respective scopes, the objective is also to ensure ownership of the overall strategy for measuring and improving the environmental footprint of IT infrastructures.