How do you measure the energy efficiency of a data center?

Data centers are big consumers of energy. Current previsions indicate than in a few years, IT system energy usage will reach almost 25% of total energy usage worldwide. Data centers are being built with energetic capacities of up to 60MW right now. For a point of comparison, consider that the entire community of Cantabria, Spain has access to 35MW.

In a climate where energy is scarce and expensive and there’s a need to reduce CO2, the majority of data centers have implemented optimization strategies for energy use. The goal is to stay competitive in the market while maintaining a commitment to the environment, which is becoming a bigger and bigger challenge that requires constant innovation.

How is energy efficiency calculated for a data center?

To calculate the energy efficiency of a data center a measure called PUE, power usage effectiveness, is used. Initially developed by a consortium called The Green Grid in 2007, PUE has been a global ISO standard since 2016.

Cooling systems, combined with the energy needed to feed the data center’s equipment, are two of the fundamental factors that determine, in their majority, the costs of energy consumption in data centers.

How is the PUE of a data center calculated?

PUE is calculated by dividing the total energy that comes into the data center by the energy used specifically by the IT equipment inside the data center. This allows us to find out the volume of energy that a data center destines to the functioning of IT systems and contrast that with the volume of energy necessary for the cooling of that equipment. The lower the result of this operation, the higher the efficiency of the data center.

The following table demonstrates an example of the relationship between PUE and energy efficiency. The ideal PUE, an impossible feat, has a value of 1. But more typically, PUE value is between 1.2 and 2.5. Adam’s data centers presently have an approximate PUE between 1.2 and 1.3, principally thanks to the free-cooling systems implemented in our data centers, the geographical location selected and the modular chamber design which allows us to periodically put into use all of the technological advances offered by our equipment providers.

PUE

Efficiency level

3.0

Very inefficient

2.5

Inefficient

2.0

Normal

1.5

Efficient

1.2

Very Efficient

*Energy efficiency levels for data centers according to The Green Grid.

What variables affect PUE in a data center?

Taking into account the continued increase in energy costs, and the ecological impact of data centers, the majority of builders and managers are centering more and more of their attention on the optimization of efficient energy use. A series of variables can be managed and implemented to improve energy efficiency such as:

  • The data center design: The use of modular chamber designs allows for growth, and adopting new technologies as they come out, as equipment constantly evolves and improves with each new version. Modular designs using cold-containment aisles also allow optimal and rational energy use.
  • The use of new technologies: The use of new systems like adiabatic cooling and free-cooling, allow the use of natural cooling methods, which reduces energy use.
  • Physical location: Obviously, the colder it is outside, in the case of systems using free-cooling, the more natural cooling can be used to control the temperature inside the data center.
  • Installation usage rate: Providing rooms with controlled density and avoiding hot spots, allows cooling systems to be optimized.
  • The equipment’s energy efficiency: The equipment, including the UPS and most modern chillers, can manage larger volumes of work while reducing energy use. A data center with more modern equipment will be more energy efficient than one that uses older equipment.

How to increase energy efficiency in a data center

There are a series of improvements to the installations, that once applied can sensibly improve final results, as far as efficiency:

  • Measurement: Without measurement, there’s no possibility of improvement. To evaluate any improvement in efficiency, it’s necessary to start by measuring existing energy usage over time. This makes it possible to compare different points in time to understand how changes affect the data center’s energy efficiency.
  • Separation of airflow: the control of currents of air, the confinement and separation of them with aisles, is crucial to reducing PUE. The hot air expelled by the equipment and the flow of cold air that comes in through the cooling systems have to be managed separately. To the contrary, the mix of these different temperature airs drastically reduces the effectivity of the cooling system. The placement and separation of cold and hot aisles, with the front part of the servers on one side, permits the creation of cold-containment aisles directing cold air to the front of the servers while hot air is directed to the back of the equipment in so-called hot aisles.
  • Check the technical floors: the technical floors are elements to consider to avoid possible air leaks. It’s essential that the floors be constantly reviewed, as since they’re made up of adjustable, moveable tiles, and provide pressurized cold air with multiple potential escape routes. This type of floor is also susceptible to being handled incorrectly, increasing the possibility of poor airflow distribution, or in the worst-case scenario, a total loss of effectivity.
  • Control the air temperature: The higher the temperature that the data center functions at, the higher its efficiency, as it needs less capacity for cooling. Operating in the mid-to-high-temperature ranges as indicated by ASHRAE standards provides additional energy efficiency. Dressing for the arctic to visit data centers is a thing of the past.
  • Humidity regulation: managing moisture in the environment can improve a data center’s efficiency. Operating with slightly higher levels of humidity within the standard and the SLAs, reduces the need for humidification and dehumidification, which produces corresponding energy savings.
  • Modular chambers: building a modular data processing system allows the periodic and gradual application of innovations that appear on the market. Many of these improvements are focused on optimizing resources and improving energy efficiency.

At Adam, with 30 years of experience, 18 of them in the management and the construction of 4 of our own data centers, we’ve learned planning for capacity in data centers isn’t a simple task. Even the best-thought-out previsions are susceptible to change, given that the sector is in a state of constant flux, with technological innovations appearing every few months. In today’s IT sector, trying to plan long-term strategies is a highly complex challenge.

Nevertheless, for some challenges, it’s worth making the effort. Without a doubt, energy efficiency is one of those challenges and a high-ranking priority for today’s operational management and business management teams.