Home
blog
data center and carbon footprint

IT Monitoring

How optimization of data centers can reduce carbon footprint

Q: What is PUE, and why does it matter?

Power Usage Effectiveness (PUE) is the total energy a facility uses divided by the energy that reaches the IT equipment. A value of 1.0 would mean no overhead at all. Modern efficient facilities typically reach roughly 1.1 to 1.3, while older ones can be well above 1.6. It is a useful efficiency indicator, but it says nothing about how clean the electricity is or how well the servers themselves are used.

Released on

Tuesday, February 28, 2023

Nikita Belov

How optimization of data centers can reduce carbon footprint

4:45

In the past 10 years, the capacity of the world's data centers has more than quadrupled. Servers can store and process more than 8 zettabytes globally as of today. It accelerates digital transformation and new business opportunities. However, it consumes a tremendous amount of energy. This blog discusses the environmental impact of large data centers and best practices to use those resources more efficiently.

Why are data centers environmentally relevant?

The ICT technology (servers, networks, terminals) produces from 2,5% to 3,7% of global CO2 emissions, putting it in the same range as commercial aviation (3,5%). In the United States, data centers account for nearly 1.8% of electricity use. The volume of emissions depends on electricity demand (to run the servers), water consumption (to cool the servers), and the lifetime of the physical servers (impact through replacements).

There are several different factors that impact the consumption of resources:

1. Deployment method (cloud/hybrid/on-premises)

The larger a data center is, the more efficiently it typically operates. That is why the CO2 footprint per square meter of hyperscale cloud providers like AWS, Azure or Google Cloud is 5 times smaller than an internal data center. The same logic applies to water consumption. Hyperscale data centers are 6 times more water-efficient compared to internal ones. As for amortization, the fewer devices you have, the less hardware you need to repair or replace. As a rule, a server should be replaced every 3 to 5 years (depending on how intensively it's used), so in the end, purchasing fewer devices not only cuts costs but also reduces environmentally harmful emissions.

2. Energy consumption per server

It is important to precisely measure the energy consumption of your servers. On the hardware side, best practice is to use highly efficient components (e.g., SSDs instead of HDDs). On the software side, virtualization can be a driver of energy efficiency. For instance, VMware claims that every virtualized server saves up to 7,000 kWh electricity per year. Instead of operating many servers at low utilization, virtualization combines the processing power onto fewer servers that operate much more efficiently.

Moreover, you must check for any idle or underutilized servers that can be turned off, repurposed, or optimized. To do so, you need an overview of what software you are using. Large organizations often tend to over-purchase software licenses leading to ineffective or unnecessary software usage within the data centers. As a result, you need more servers, VMs, storage, and energy to power it all. By reducing or optimizing software usage, you can decrease the number of devices and reduce the environmental impact of your IT infrastructure. USU Discovery provides you with a detailed and real-time overview of what devices and software your company is using.

3. Electricity source

The availability of different energy sources for your data centers depends on their location. Using renewable energy providers can significantly reduce your emissions. If your data center is on the cloud, you can use the provider's measurement tools or documentation to estimate the ecological impact. Cloud providers vary in their investments in carbon-free energy depending on their location. So, if you want to reduce your carbon footprint, you can select to host in a more environmentally friendly region.

usu_wp_cloud-discovery_en_cover_800x800px

The Data Discovery Pain in the As-a-Service

What you can do right now

There are some simple things you can do right away to reduce your server usage and cut your carbon footprint:

Switch to environmentally friendly electricity providers – it will not only reduce your carbon footprint, but also decrease air and water pollution, protect natural habitats, and support the development of sustainable communities.
Conduct a thorough inventory of your IT environment - this will allow you to identify where excess servers have been installed and assess whether they’re necessary.
Implement a reduction strategy – reduce the number of servers in your IT environment by either depreciating underused servers, using more efficient physical and virtual technologies, or moving to a public cloud.
Avoid redundant software - using only relevant software helps to improve energy efficiency, reduce resource waste, and eventually decrease your carbon footprint.

Your organization (still) lacks precise visibility of what hardware devices and software are running? Let our SAM solution shed some light and ask for a demo now.

Frequently Asked Questions

Text Link

Why do data centers have a carbon footprint?

Data centers run around the clock and draw a lot of electricity for servers, storage, networking and cooling. Emissions come from the power generation behind that electricity, from the manufacturing and transport of hardware, and from construction of the facility itself. How large the footprint is depends heavily on the local energy mix.

What is PUE, and why does it matter?

Power Usage Effectiveness (PUE) is the total energy a facility uses divided by the energy that reaches the IT equipment. A value of 1.0 would mean no overhead at all. Modern efficient facilities typically reach roughly 1.1 to 1.3, while older ones can be well above 1.6. It is a useful efficiency indicator, but it says nothing about how clean the electricity is or how well the servers themselves are used.

What are Scope 1, 2 and 3 emissions in a data center context?

Scope 1 covers direct emissions, such as diesel backup generators or refrigerant leaks. Scope 2 covers emissions from purchased electricity and cooling. Scope 3 covers everything else in the value chain, including the production of servers and chips, construction materials, and the end-of-life handling of equipment. For many operators, Scope 2 dominates, but Scope 3 is growing in importance as grids get cleaner.

Does moving workloads to the cloud automatically reduce emissions?

Not automatically. Large cloud providers often run more efficient infrastructure and higher server utilization than a typical private data center. But migrating wasteful workloads simply moves the waste, and unmanaged cloud usage can lead to oversized instances and forgotten resources. Emissions only drop if you also right-size and clean up what you run.

What is an ITSM tool?

An ITSM tool streamlines and automates IT services, helping you efficiently manage support tickets, changes, problem resolution, and asset tracking. Our adaptable solution goes beyond IT, supporting other service areas across your organization.

What is the difference between location-based and market-based carbon accounting?

The location-based method uses the average emissions of the local grid where the electricity is consumed. The market-based method reflects the energy purchases an organization has made, such as renewable energy certificates or power purchase agreements. Reporting both gives a more honest picture, because buying certificates does not always mean the grid at that moment and place was actually clean.

How can cooling be made more sustainable?

Cooling can account for a significant share of a facility's energy use. Common measures include hot/cold aisle containment, raising the allowed operating temperature, free cooling with outside air in cooler climates, and liquid cooling for high-density racks. Reusing waste heat, for example to supply district heating, can offset emissions elsewhere. Water use is a trade-off, since some efficient cooling methods consume a lot of water

Why does AI make data center energy use a bigger topic?

AI training and inference rely on GPUs and accelerators that draw far more power per rack than conventional servers. This pushes up both electricity demand and cooling requirements, and it often concentrates load in specific regions. Efficiency gains in hardware and models help, but overall demand is still expected to rise as AI usage grows.

How long should servers be kept, from a carbon perspective?

Manufacturing a server causes a significant share of its lifetime emissions, so replacing hardware too early wastes embodied carbon. Keeping it too long can also be counterproductive if newer models do the same work with much less energy. The best refresh cycle depends on the workload, the efficiency gain of new hardware, and the carbon intensity of the electricity used. Reuse, refurbishment and certified recycling reduce the impact further.

Which regulations and standards apply to data center sustainability?

In the EU, the Energy Efficiency Directive requires larger data centers to report energy and water performance, and the Corporate Sustainability Reporting Directive (CSRD) pushes companies to disclose their emissions, including IT-related ones. Germany has its own Energy Efficiency Act with specific requirements for data centers. Voluntary frameworks such as the GHG Protocol, ISO 14001, ISO 50001 and the EU Code of Conduct for Data Centres help organizations structure their measurement and improvement efforts. Requirements and thresholds change, so check the current legal texts before citing details.