Data Center

Data Center

Introduction to Power Usage Effectiveness in Data Centers

Data Center

Introduction to Power Usage Effectiveness in Data Centers

By

Dr. Sudhakar Gummadi

Aug 14, 2024

10

min read

The regulatory and compliance requirements for data centers are getting stricter with every passing day. We all know how much of a role Power Usage Effectiveness or PUE plays in all these regulations and compliance policies.

Even though it’s not the only efficiency metric out there, it’s essential. For everyone aiming for lower PUE values, the challenges are endless. And finding the right balance between efficiency and performance is a challenge.

Climate, geography, IT infrastructure, design, and cooling - all play their parts in affecting the PUE values.

Here’s when a partner like Arche, with 17+ years of experience, can help you with your data center PUE values. In this blog, we’ll uncover what PUE is, how we calculate it, and understand why it’s a defining factor for any data center. We’ll also understand how certain factors affect the PUE of your data centers.

Want to lower the PUE value of your data center? Contact our data center team to know how they can help.

What is PUE and Why is it Important?

Efficiency for any system needs to be the most significant criterion. However capable and powerful a system is, it’s useless if it isn’t efficient. For data centers, one of the ways to calculate efficiency is by using Power Usage Effectiveness or PUE.

Defining PUE in Data Centers

Power Usage Effectiveness is the ratio of the total power used by the entire facility and the total power used by all the IT equipment in the data center.

Commonly, facilities use it as a metric to measure their efficiency. The closer it is to 1, the more efficient the data center facility is.

Understanding PUE Metrics and Their Importance in Energy Efficiency

PUE as a metric is vital. You cannot improve what you cannot measure.

When we talk about the total power used by the facility, it includes everything that runs on power within the facility. Lights, cooling, HVAC, backup, and other ancillary systems use power.

The focus is always on the power used by the servers, the storage, and the racks within the data center infrastructure. Facilities should aim to direct the maximum power towards these core data center elements to get the maximum out of their facilities.

Here’s where the PUE metric helps data center facilities be more aware of their energy efficiency. It helps them become aware of several aspects.

Energy Efficiency

Although it isn’t the most accurate method to measure efficiency, it gives you a pretty good idea of how efficient your facility is. It gives you a clear number, a ratio of total power used vs power used by core systems.

And that urges you to take action. It pushes the facilities to find ways to be more efficient.

Costs

Knowing that you need to lower your energy consumption will automatically reduce the costs. Even if you don’t lower the overall energy consumption, the ratio of total power and the power going to the core system will go up.

That means you’ll get better returns on your investment.

Environmental Impact

Let’s say your PUE is closer to 1 than before. And your performance requirements are the same. So, the energy efficiency is higher, you’re doing the same amount of work while using less energy, and the costs are also lower.

All these will result in lower environmental impact as your facility still gives the same output as before while using less power.

Operational Efficiency

As the overall efficiency goes up, your operational efficiency will go up. That’s the only way to get your PUE closer to 1.

Better power management of all the systems, core as well as ancillary, will boost your overall efficiency, leading to better power usage, better hardware and software optimizations, and overall reliability of the facility.

Regulatory Compliance

Several regions and mandates, most of them voluntary, push the data centers to be more efficient and responsible towards the environment.

  1. The European Union Code of Conduct for Data Centers: It’s a voluntary initiative by the EU, and it aims to bring the PUE of data centers in Europe as close to 1 as possible.

  2. The Energy Efficiency Directive: It’ll make it mandatory for all the data centers larger than 500kW in Europe to report their metrics, PUE being one of them, from May 2024.

  3. Indian Green Building Council Data Centre Ratings: Based on your PUE, IGBC awards points to the data centers. It's voluntary, and data centers need not always get points from IGBC.

    IGBC gives 20 points to a data center when its PUE is between 1.2 and 1.4. And it offers 42 points to any data center that manages to get its PUE less than 1.1.

  4. Germany’s Energy Efficiency Act: It requires all new data centers that’ll be operational after 2026 to have a PUE of 1.2 or lower within two years of commissioning. By 2030, all data centers must operate with a PUE of 1.3 or lower.

There are many such regulations across the globe, most of them voluntary. But with increasing energy consumption and focus on sustainability, expect stricter mandatory regulations like Germany’s Energy Efficiency Act.

All these regulations and guidelines underscore the importance of the PUE as a metric for data centers.

Common PUE Levels and Standards

Okay, so I’ve thrown a lot of PUE values in the above sections. We know that a PUE value closer to 1 is better. But the real world is quite different from that.

What is a Good PUE for a Data Center?

If we go back to the definition of Power Usage Effectiveness or PUE, it’s the ratio of the total power used by the facility and the power used by the core IT equipment. That core IT equipment can vary depending on the data center.

Using this definition, when the PUE of a data center is, let’s say, 1.5, it’s using 1.5 times the energy used by its core IT equipment to keep the entire facility running.

In an ideal world, data centers would have a PUE value of 1. We’ve discussed that before. But, that’s only possible theoretically. In the practical world, a lot of losses occur. You'll need cooling and lighting systems to keep the data center at optimum operating temperature.

Depending on the various external and internal factors, the PUE can vary for data centers across the globe.

Internal factors such as the design of the cooling systems, the layout of the data center, free cooling space, etc affect the PUE. Radical designs and efficient cooling systems push the PUE closer to 1.

But you cannot change the external factors. However cheeky you get with optimizing the internal factors, you cannot change the external factors like weather and geographical conditions.

If a data center facility sits in the Indian subcontinent, the weather WILL be hot and humid. You cannot change that. And to compensate for that, you’ll need extra cooling. That’ll push the PUE farther from 1.

But if you construct a data center near the Arctic Circle, you won’t need as much cooling as the one in the Indian subcontinent. Your PUE will be much closer to 1.

So, depending on the geography and climate, the PUE will vary. Let’s look at what ideal PUE values across industries.

If you're familiar with PUE and want to optimize it for better efficiency, you'll find our blog on it to be insightful.

Average and Ideal PUE Values Across the Industry

An ideal PUE value of 1 isn’t practically possible. If that were possible, all the ancillary systems would consume zero power. But no data center can achieve that. Let’s look at the common PUE values and what they mean.

  • 1.0: Theoretically perfect efficiency, but not achievable in practice

  • 1.1-1.2: Highly efficient data centers using the best practices and innovative technologies

  • 1.2-1.4: Well-designed new data centers using efficient cooling and power distribution

  • 1.4-1.5: Efficiently designed and operated data centers

  • 1.5-1.7: Average data centers using conventional cooling and power infrastructure

  • 1.7-2.0: Older, inefficient data centers with significant overhead

  • 2.0+: Very inefficient data centers with high overhead

From these values, we can draw several conclusions.

  • New data centers should aim for PUE values of around 1.2.

  • Older data centers can optimize their systems to maintain PUE values of 1.5 or less.

  • Hyperscalar data centers can achieve lower PUE values because of the scale at which they operate.

  • Data centers in warm climates, as we saw in the example above, will traditionally always have higher PUE values.

Factors Influencing PUE

PUE is not a constant. It varies frequently. That’s why facilities usually measure it at regular intervals. These intervals can be monthly, weekly, daily, or even hourly.

However, the reasons behind the fluctuations in the PUE value are many. Let’s look at some of them.

How Cooling and Energy Use Impact PUE

Cooling

Multiple racks in the data center that contain servers, storage units, and more generate enormous amounts of heat. If that heat isn’t disposed of properly, they won’t give you the performance you expect from them.

Here’s where cooling comes into the picture. Traditionally, cooling systems account for a big chunk of the total energy consumption, usually between 30-50%. This large chunk of energy spent on cooling directly increases the value of PUE.

But with modern cooling systems like liquid cooling, free cooling, and smart racks, the facilities can reduce their cooling energy spend by 10-20%.

Energy Usage

An outdated, end-of-life server or a rack in your data center WILL eat up more energy than a modern, optimized smart rack.

If you compare the performance per watt of the servers from 2008-09 with the servers from 2018-19 under load, the number has gone up by almost 8 times according to a report by IEEE.

Outdated hardware is single-handedly pulling your PUE values down.

Modern hardware, when combined with renewable energy sources like solar or wind energy will help your data center achieve more performance output while consuming less power.

The Role of Data Center Design in Achieving Optimal PUE

If the hot air coming out of one of the racks of servers in your data center is blowing directly in front of the other rack of servers, it isn’t a great design.

You’re reducing the efficiency of the rack behind. And to bring it back to its desired efficiency and performance output, you’ll need more cooling. More cooling means more power consumption. And more power consumption equals a higher PUE value.

Airflow Management

The above example explains to you the importance of airflow management. Several techniques help you achieve better PUE values.

Techniques like hot/cold aisle containment, blanking panels, and cable routing optimizations can help the airflow, enabling better cooling and higher efficiency. You’ll need less cooling power.

Data Centre Design

Data center footprint, rack density and placement, and equipment positioning when done to optimise airflow and cooling can save you energy that goes into cooling. It can help reduce power losses and eventually better PUE value.

Over-provisioning of data center resources, when not needed, increases idle power consumption. Having a modular, scalable data center design can help you minimize wastage of resources. A smaller number of active servers equals less heat and power to cool the facility.

All these techniques can help you achieve lower PUE.

How Arche AI Can Help

Power Usage Effectiveness in modern data center operations is a metric you cannot ignore. Our team of experts specializes in optimizing data center efficiency, helping you achieve lower PUE values and drive sustainable, cost-effective operations.

We offer comprehensive data center assessment services, analyzing your current infrastructure, cooling systems, and energy usage patterns. Our consultants will work closely with you to identify areas for improvement and develop tailored strategies to enhance your facility's energy efficiency.

Arche's solutions encompass:

  • Advanced cooling system design and implementation

  • Airflow management optimization

  • Energy-efficient hardware recommendations

  • Data center layout and design enhancements

  • Continuous monitoring and reporting of PUE metrics

Our approach ensures that your data center not only meets industry standards but also achieves optimal performance while minimizing environmental impact.

Don't let inefficiencies drain your resources. Arche can help you unlock the full potential of your data center's energy efficiency. Contact us today to schedule a consultation and take the first step towards a more sustainable, cost-effective data center operation.

Table of contents

POINTS

POINTS

POINTS

Share artilce:

linkedin
linkedin
twitter
twitter

Written by

Dr. Sudhakar Gummadi

Director – Consulting Services

Dr. Sudhakar Gummadi has a Doctorate Degree and 35+ years' expertise to drive business innovation. Specializing in information security, networking, and risk management, he excels in multiplatform enterprise security, compliance, and data centers. His contributions extend to industrial automation, laboratory installations, and teaching post-graduate level courses in Network Engineering, Network Security, and more.

Linkedin
Linkedin

Ready to take your company to the next level?

Transformation starts here, talk to our experts

Ready to take your company to the next level?

Transformation starts here, talk to our experts

Ready to take your company to the next level?

Transformation starts here, talk to our experts

Ready to take your company to the next level?

Transformation starts here, talk to our experts

BLOG

Introduction to Power Usage Effectiveness in Data Centers

BY

Dr. Sudhakar Gummadi

Aug 14, 2024

10

min read

The regulatory and compliance requirements for data centers are getting stricter with every passing day. We all know how much of a role Power Usage Effectiveness or PUE plays in all these regulations and compliance policies.

Even though it’s not the only efficiency metric out there, it’s essential. For everyone aiming for lower PUE values, the challenges are endless. And finding the right balance between efficiency and performance is a challenge.

Climate, geography, IT infrastructure, design, and cooling - all play their parts in affecting the PUE values.

Here’s when a partner like Arche, with 17+ years of experience, can help you with your data center PUE values. In this blog, we’ll uncover what PUE is, how we calculate it, and understand why it’s a defining factor for any data center. We’ll also understand how certain factors affect the PUE of your data centers.

Want to lower the PUE value of your data center? Contact our data center team to know how they can help.

What is PUE and Why is it Important?

Efficiency for any system needs to be the most significant criterion. However capable and powerful a system is, it’s useless if it isn’t efficient. For data centers, one of the ways to calculate efficiency is by using Power Usage Effectiveness or PUE.

Defining PUE in Data Centers

Power Usage Effectiveness is the ratio of the total power used by the entire facility and the total power used by all the IT equipment in the data center.

Commonly, facilities use it as a metric to measure their efficiency. The closer it is to 1, the more efficient the data center facility is.

Understanding PUE Metrics and Their Importance in Energy Efficiency

PUE as a metric is vital. You cannot improve what you cannot measure.

When we talk about the total power used by the facility, it includes everything that runs on power within the facility. Lights, cooling, HVAC, backup, and other ancillary systems use power.

The focus is always on the power used by the servers, the storage, and the racks within the data center infrastructure. Facilities should aim to direct the maximum power towards these core data center elements to get the maximum out of their facilities.

Here’s where the PUE metric helps data center facilities be more aware of their energy efficiency. It helps them become aware of several aspects.

Energy Efficiency

Although it isn’t the most accurate method to measure efficiency, it gives you a pretty good idea of how efficient your facility is. It gives you a clear number, a ratio of total power used vs power used by core systems.

And that urges you to take action. It pushes the facilities to find ways to be more efficient.

Costs

Knowing that you need to lower your energy consumption will automatically reduce the costs. Even if you don’t lower the overall energy consumption, the ratio of total power and the power going to the core system will go up.

That means you’ll get better returns on your investment.

Environmental Impact

Let’s say your PUE is closer to 1 than before. And your performance requirements are the same. So, the energy efficiency is higher, you’re doing the same amount of work while using less energy, and the costs are also lower.

All these will result in lower environmental impact as your facility still gives the same output as before while using less power.

Operational Efficiency

As the overall efficiency goes up, your operational efficiency will go up. That’s the only way to get your PUE closer to 1.

Better power management of all the systems, core as well as ancillary, will boost your overall efficiency, leading to better power usage, better hardware and software optimizations, and overall reliability of the facility.

Regulatory Compliance

Several regions and mandates, most of them voluntary, push the data centers to be more efficient and responsible towards the environment.

  1. The European Union Code of Conduct for Data Centers: It’s a voluntary initiative by the EU, and it aims to bring the PUE of data centers in Europe as close to 1 as possible.

  2. The Energy Efficiency Directive: It’ll make it mandatory for all the data centers larger than 500kW in Europe to report their metrics, PUE being one of them, from May 2024.

  3. Indian Green Building Council Data Centre Ratings: Based on your PUE, IGBC awards points to the data centers. It's voluntary, and data centers need not always get points from IGBC.

    IGBC gives 20 points to a data center when its PUE is between 1.2 and 1.4. And it offers 42 points to any data center that manages to get its PUE less than 1.1.

  4. Germany’s Energy Efficiency Act: It requires all new data centers that’ll be operational after 2026 to have a PUE of 1.2 or lower within two years of commissioning. By 2030, all data centers must operate with a PUE of 1.3 or lower.

There are many such regulations across the globe, most of them voluntary. But with increasing energy consumption and focus on sustainability, expect stricter mandatory regulations like Germany’s Energy Efficiency Act.

All these regulations and guidelines underscore the importance of the PUE as a metric for data centers.

Common PUE Levels and Standards

Okay, so I’ve thrown a lot of PUE values in the above sections. We know that a PUE value closer to 1 is better. But the real world is quite different from that.

What is a Good PUE for a Data Center?

If we go back to the definition of Power Usage Effectiveness or PUE, it’s the ratio of the total power used by the facility and the power used by the core IT equipment. That core IT equipment can vary depending on the data center.

Using this definition, when the PUE of a data center is, let’s say, 1.5, it’s using 1.5 times the energy used by its core IT equipment to keep the entire facility running.

In an ideal world, data centers would have a PUE value of 1. We’ve discussed that before. But, that’s only possible theoretically. In the practical world, a lot of losses occur. You'll need cooling and lighting systems to keep the data center at optimum operating temperature.

Depending on the various external and internal factors, the PUE can vary for data centers across the globe.

Internal factors such as the design of the cooling systems, the layout of the data center, free cooling space, etc affect the PUE. Radical designs and efficient cooling systems push the PUE closer to 1.

But you cannot change the external factors. However cheeky you get with optimizing the internal factors, you cannot change the external factors like weather and geographical conditions.

If a data center facility sits in the Indian subcontinent, the weather WILL be hot and humid. You cannot change that. And to compensate for that, you’ll need extra cooling. That’ll push the PUE farther from 1.

But if you construct a data center near the Arctic Circle, you won’t need as much cooling as the one in the Indian subcontinent. Your PUE will be much closer to 1.

So, depending on the geography and climate, the PUE will vary. Let’s look at what ideal PUE values across industries.

If you're familiar with PUE and want to optimize it for better efficiency, you'll find our blog on it to be insightful.

Average and Ideal PUE Values Across the Industry

An ideal PUE value of 1 isn’t practically possible. If that were possible, all the ancillary systems would consume zero power. But no data center can achieve that. Let’s look at the common PUE values and what they mean.

  • 1.0: Theoretically perfect efficiency, but not achievable in practice

  • 1.1-1.2: Highly efficient data centers using the best practices and innovative technologies

  • 1.2-1.4: Well-designed new data centers using efficient cooling and power distribution

  • 1.4-1.5: Efficiently designed and operated data centers

  • 1.5-1.7: Average data centers using conventional cooling and power infrastructure

  • 1.7-2.0: Older, inefficient data centers with significant overhead

  • 2.0+: Very inefficient data centers with high overhead

From these values, we can draw several conclusions.

  • New data centers should aim for PUE values of around 1.2.

  • Older data centers can optimize their systems to maintain PUE values of 1.5 or less.

  • Hyperscalar data centers can achieve lower PUE values because of the scale at which they operate.

  • Data centers in warm climates, as we saw in the example above, will traditionally always have higher PUE values.

Factors Influencing PUE

PUE is not a constant. It varies frequently. That’s why facilities usually measure it at regular intervals. These intervals can be monthly, weekly, daily, or even hourly.

However, the reasons behind the fluctuations in the PUE value are many. Let’s look at some of them.

How Cooling and Energy Use Impact PUE

Cooling

Multiple racks in the data center that contain servers, storage units, and more generate enormous amounts of heat. If that heat isn’t disposed of properly, they won’t give you the performance you expect from them.

Here’s where cooling comes into the picture. Traditionally, cooling systems account for a big chunk of the total energy consumption, usually between 30-50%. This large chunk of energy spent on cooling directly increases the value of PUE.

But with modern cooling systems like liquid cooling, free cooling, and smart racks, the facilities can reduce their cooling energy spend by 10-20%.

Energy Usage

An outdated, end-of-life server or a rack in your data center WILL eat up more energy than a modern, optimized smart rack.

If you compare the performance per watt of the servers from 2008-09 with the servers from 2018-19 under load, the number has gone up by almost 8 times according to a report by IEEE.

Outdated hardware is single-handedly pulling your PUE values down.

Modern hardware, when combined with renewable energy sources like solar or wind energy will help your data center achieve more performance output while consuming less power.

The Role of Data Center Design in Achieving Optimal PUE

If the hot air coming out of one of the racks of servers in your data center is blowing directly in front of the other rack of servers, it isn’t a great design.

You’re reducing the efficiency of the rack behind. And to bring it back to its desired efficiency and performance output, you’ll need more cooling. More cooling means more power consumption. And more power consumption equals a higher PUE value.

Airflow Management

The above example explains to you the importance of airflow management. Several techniques help you achieve better PUE values.

Techniques like hot/cold aisle containment, blanking panels, and cable routing optimizations can help the airflow, enabling better cooling and higher efficiency. You’ll need less cooling power.

Data Centre Design

Data center footprint, rack density and placement, and equipment positioning when done to optimise airflow and cooling can save you energy that goes into cooling. It can help reduce power losses and eventually better PUE value.

Over-provisioning of data center resources, when not needed, increases idle power consumption. Having a modular, scalable data center design can help you minimize wastage of resources. A smaller number of active servers equals less heat and power to cool the facility.

All these techniques can help you achieve lower PUE.

How Arche AI Can Help

Power Usage Effectiveness in modern data center operations is a metric you cannot ignore. Our team of experts specializes in optimizing data center efficiency, helping you achieve lower PUE values and drive sustainable, cost-effective operations.

We offer comprehensive data center assessment services, analyzing your current infrastructure, cooling systems, and energy usage patterns. Our consultants will work closely with you to identify areas for improvement and develop tailored strategies to enhance your facility's energy efficiency.

Arche's solutions encompass:

  • Advanced cooling system design and implementation

  • Airflow management optimization

  • Energy-efficient hardware recommendations

  • Data center layout and design enhancements

  • Continuous monitoring and reporting of PUE metrics

Our approach ensures that your data center not only meets industry standards but also achieves optimal performance while minimizing environmental impact.

Don't let inefficiencies drain your resources. Arche can help you unlock the full potential of your data center's energy efficiency. Contact us today to schedule a consultation and take the first step towards a more sustainable, cost-effective data center operation.

Linkedin

Written by

Dr. Sudhakar Gummadi

Director – Consulting Services

Dr. Sudhakar Gummadi has a Doctorate Degree and 35+ years' expertise to drive business innovation. Specializing in information security, networking, and risk management, he excels in multiplatform enterprise security, compliance, and data centers. His contributions extend to industrial automation, laboratory installations, and teaching post-graduate level courses in Network Engineering, Network Security, and more.

Share artilce:

linkedin
twitter

Partner with us

Unlock your business potential with our committed team driving your success.

Data Center

Ready to take your company to the next level?

Transformation starts here, talk to our experts

Ready to take your company to the next level?

Transformation starts here, talk to our experts