What is PUE? And how is it related to sustainability?

Learn about PUE, the metric that drives sustainability efforts, addressing the challenge of meeting the growing demand for data in an eco-friendly way.

The increasing reliance on the digital world brings an environmental challenge into focus. For processing and storing large volumes of data, we are seeing more and more data centers being built, whether centralized or at the edge, significantly increasing the planet's energy consumption.

According to the International Energy Agency (IEA), energy consumption in data centers is set to double between 2022 and 2026, driven mainly by new technologies like AI and cryptocurrencies. Both applications require processing large volumes of data, which in turn leads to high energy consumption.

Aware of this variable, the Green Grid introduced to the industry an indicator that has become the most widely used to define the energy efficiency of data facilities: Data Center Power Usage Effectiveness (PUE).

Since its introduction, PUE has helped focus attention on the efficiency of mechanical and electrical infrastructure, one of the main causes of energy waste and inefficiency in data centers, and has driven the adoption of many recommended energy and cooling practices.

Is an efficient PUE high or low?

According to this metric, a low PUE indicates energy efficiency, reduced operational costs, and lower environmental impact. A high PUE, on the other hand, signals energy waste, high operational costs, and significant environmental impacts.

It is essential to analyze that the cooling system of the data center installation is the central piece to define effective strategies capable of improving PUE. Current trends even consider the possibility of using liquid cooling, in addition to adopting renewable energy sources to reduce costs, maximize performance, prioritize sustainability, and make operations even more efficient.

How is PUE calculated in a data center?

PUE compares the total energy consumption of all components, including cooling systems, lighting, security, and auxiliary equipment, divided by the energy consumption of IT equipment responsible for processing and storing data. The ideal result should be close to 1.

Edgefy: como se calcula o PUE em um data center

Figure 1: Calculation of Data Center Power Usage Effectiveness (PUE)

The Data Center Power Usage Effectiveness (PUE) metric is important for identifying opportunities to improve the operational efficiency of data centers, compare similar facilities, implement project and process improvements , and adjust energy for additional IT equipment.

How is a PUE result classified?

Given that PUE is the ratio between the total energy consumption of the facilities and the energy consumption of IT equipment, it is possible to compare the efficiency of the data center using a benchmark table that classifies results from Platinum (PUE lower than 1.25) to Unrecognized (PUE greater than 2.5), passing through other levels like Gold (PUE between 1.35 and 1.43), Silver (PUE between 1.43 and 1.67), and Bronze (PUE between 1.67 and 2).

Check the explanatory chart:

Edgefy: Como se classifica o resultado de um PUE?

Figure 2: Benchmark of Data Center Power Usage Effectiveness (PUE)

The closer the facilities are to a Platinum PUE, the less energy is consumed by cooling systems and other auxiliary equipment. Considering that data centers require electricity not only to run 24/7 but also for their complex cooling systems, the goal is always to aim for a Platinum PUE, either during installation or in opportunities to review projects and expand data storage activities.

Are there data on the energy efficiency of data centers?

According to Google’s 2022 Sustainability Report, cloud storage capacity is now seven times larger than it was five years ago. One of the major challenges for the data center market is reducing energy consumption to contribute to environmental and business goals so that the data industry can evolve by prioritizing energy efficiency and reducing the use of heavy metals..

Since 2018, the Uptime Institute has warned about the likelihood of stricter government regulations and increased public pressure regarding the sustainability of data center facilities.

According to Uptime Institute analyses, by 2030, as organizations fail to meet sustainability goals and legal requirements, penalties are expected to become more stringent, frequent, and costly from both a financial and operational perspective.

Improvements are expected as the reality of obtaining clean or renewable energy becomes more attainable for the data center industry. Best practices, such as separating hot and cold air or building entire data centers designed to minimize energy inefficiencies, in addition to replacing older and less efficient cooling systems with newer ones, already represent a significant global reduction in PUE, as shown by the Uptime Institute’s chart, from 2023.

Average annual PUE of Data Centers (n=567)

Edgefy: O que é PUE? - Pue anual médio de Data Centers (n=567)

Figure 3: Energy Efficiency Gains Have Stagnated

However, the data shows some stagnation in energy efficiency gains in recent years. It seems that the best practices implemented in the early years of the PUE metric were those that did not require substantial investments but rather process changes with relatively small and incremental financial inputs. See the list of best practices for PUE reduction, compiled by one of with relatively small and incremental financial inputs. See the list of best practices for PUE reduction, compiled by one of:

Edgefy: o que é PUE? principais práticas reconhecidas pelo código de conduta da comissão européia

Figure 4: Most Implemented Practices for Data Center Energy Efficiency, According to the European Commission's Code of Conduct

Additional advancements will require greater investments, focus, and engagement from the IT industry with short, medium, and long-term environmental goals. The European Commission also published a list of less-implemented practices, which include structural changes in the organizational culture of IT companies along with strategic operational changes that involve investments and operational risks, as demonstrated in the figure below:

Edgefy: o que é PUE? melhores práticas do código de conduta da comissão européia

Figure 5: Least Implemented Practices for Data Center Energy Efficiency, According to the European Commission's Code of Conduct

In a broader sense, looking at the trajectory of data centers shows a positive story. Servers have observed consistent energy efficiency increases, over the last few decades, allowing infrastructure operators to continuously increase computational capacity, while reducing energy consumption through hardware upgrades.

But the opportunities for energy savings are so great that if fully implemented, they could significantly reduce the global data center energy consumption.

A data center that has been in operation for more than five years, accounting for 40% of servers in use today, represents 66% of total energy consumed, but contributes only 7% of current computational capacity, as shown in the chart:

Figure 5: Least Implemented Practices for Data Center Energy Efficiency, According to the European Commission's Code of Conduct

How is the IT industry positioned in the face of global climate goals?

The message is clear. Upgrading older servers and cooling machines will significantly reduce overall energy consumption by the data industry and while there are structures that cannot be easily replaced, there are significant opportunities to update a large portion of them.

In the coming years, companies will be required to report their carbon emissions and climate-related risks, with much less flexible standards and regulatory requirements than they have been practicing so far.

Data from the Uptime Institute shows that over the past decade, we have indeed seen substantial energy efficiency gains in IT and data centers , largely attributed to a focus on infrastructure, design, equipment, and process management improvements to promote optimization and process efficiency.

However, unfortunately, optimism must be tempered by the clear evidence that energy efficiency was driven more by equipment performance than by the adoption of efficiency practices in data centers

Contrary to expectations, but aligning with results, less than half of digital infrastructure managers interviewed in IT organizations claim that they are currently tracking any type of carbon emission data (fuel, purchased electricity, and procurement of goods and services), according to two major surveys led by the Uptime Institute in partnership with with the International Business Machines Corporation (IBM).

Why does IT leadership still seem disengaged from the planet’s decarbonization and climate safety strategies?

When thinking about pollution, the most common image in the collective unconscious might be smoke billowing from a factory chimney in a classic industrial park. However, pollution happens in places and situations that many do not even imagine. The level of national development and the growing process of data digitalization influence carbon emissions.

Climate stabilization, related to decarbonization and a just energy transition, is an extension of the core theme of the Paris Agreement, revisited annually by the signatory countries of the Climate Convention at the Conferences of the Parties (COP) aiming to limit the global temperature rise to 1.5ºC above pre-industrial levels by 2100

Climate science is sophisticated, and the Intergovernmental Panel on Climate Change (IPCC) reports indicate that if we fail to reduce emissions in the 2020s, limiting warming will be beyond our reach, and the extreme climate we are already experiencing today will seem mild by comparison.

We already know that human activities have contributed to nearly 90% of the 1.1°C warming we have experienced since the pre-industrial era, while natural forces like volcanoes or solar variations account for only 0.1°C of this increase.

Governments, companies, and investors must intensify their actions in proportion and scale to the crisis we face. To give you an idea of the impact of digitalization on climate statistics, according to the World Economic Forum data centers consume 1% of all electricity generated globally. .

Is it possible to make data centers more sustainable?

In general, data centers constantly require large amounts of energy. Combined, they account for 0.3% of all global CO2 emissions and consumed about 0.9% to 1.3% of all electricity demand in the world in 2019, according to a report published by the Capgemini Research Institute. This preliminary research indicates that 150 million tons of CO2 per year are produced by the energy needs of data centers, equivalent to the emissions generated by 35 million cars. And with information traffic doubling every four years, the challenge of clean energy use only grows.

The central issue lies in the equation between the number of CPUs (a large-scale data center can house more than 100,000 servers, each with multiple processors) and the physical capacity to cool these machines. Hardware industries are already working to launch more efficient CPUs, while radical solutions are expected from data center companies.

Taking data centers to colder countries was an option even used by Facebook, which built a data reservoir the size of six football fields just 100 km from the Arctic Circle, in Sweden

While it is efficient—the Arctic platform draws in -50ºC air using turbines similar to those of airplanes, with energy efficiency up to 40% greater than a conventional data center—in practice, this doesn’t seem feasible on a global scale.

The question of how to achieve a PUE below 1.25 (Platinum) in most facilities is still wide open and undefined. We are witnessing an exciting era, with significant competitive advantages to be gained by reaching high energy efficiencies. From now on, we’ll see major transformations as data centers are built and retrofitted with the ability to simultaneously lower costs and contribute to global climate goals, when operated at optimal capacity.

Insights

+ 150

racks

+ 1,82

MW

III

Tier

1,4

PUE

Case SERPRO

Block building, modular, alta capacidade de processamento, escalável, infraestrutura de energia eficiente

Certificação NBR/ABNT 10636

O novo data center modular do SERPRO (Serviço Federal de Processamento de Dados) possui certificação conforme a norma NBR/ABNT 10636, atendendo aos critérios estabelecidos para garantir um ambiente protegido contra diversos riscos.

Tier III

Além de uma infraestrutura física robusta e resiliente, o Data Center possui certificação Tier III pelo Uptime Institute. Dessa forma é possível realizar manutenções concorrentes sem impactar no serviço para garantir a disponibilidade.

Edgefy: Case SERPRO: usina de geradores

PUE 1,4

O data center modular do SERPRO tem um PUE de 1,4 que é altamente eficiente, consumindo apenas 40% da energia total para operações de suporte, como climatização. Esse desempenho reflete um design otimizado e sustentável, garantindo economia e alta performance operacional.

Edgefy: Data center modular SERPRO: Tela de gestão do ar-condicionado

Planta baixa simplificada

O data center modular do SERPRO é formado por múltiplos containers e oferece uma infraestrutura completa e segmentada, garantindo eficiência e segurança. Com áreas separadas para front desk, NOC (Centro de Operações de Rede), data halls, salas elétricas, e uma usina de geradores, ele permite maior controle e gestão operacional. Essa estrutura modular também proporciona escalabilidade, permitindo que novos containers sejam integrados conforme a demanda cresce, além de oferecer alta resiliência por meio da segregação de sistemas críticos e operacionais.

Gostaria de saber mais sobre o Edgefy?

Ao clicar acima, você concorda com as Políticas de privacidade e cookies deste site.

Usamos cookies para garantir a melhor experiência em nosso site. Se você continuar navegando, assumiremos que está de acordo com a nossa política de privacidade e cookies.