Optimising PUE for Enhanced Data Centre Efficiency

Power Usage Effectiveness (PUE) remains a critical metric for measuring efficiency. But for many data center operators, the journey from understanding PUE to actually optimizing it can be fraught with challenges.

How do you navigate the maze of best practices and cutting-edge innovations? What strategies can truly move the needle on your facility's efficiency? And how do you balance the pursuit of a lower PUE with the realities of operational demands and budget constraints?

In this comprehensive guide, we'll cut through the noise and provide you with actionable insights to optimize your data center's PUE. We'll explore proven strategies, from airflow management to precision cooling, and dive into the technological innovations that are reshaping the efficiency landscape.

But we won't stop there. We'll also take you behind the scenes of some of the world's most efficient data centers, unpacking the lessons you can apply to your own facility. And we'll equip you with the tools and knowledge you need to accurately calculate and monitor your PUE, ensuring you can track your progress and identify areas for improvement.

Whether you're looking to make incremental gains or aiming for a dramatic efficiency overhaul, this guide will provide you with the roadmap you need to enhance your data center's performance and reduce your operational costs.

Ready to take your data center efficiency to the next level? Schedule a call with our experts to learn more.

Strategies for Reducing PUE

We covered why PUE as a metric for data centres is vital in the previous blog. But we didn’t take a deep dive into the strategies that can help you reduce the PUE for your facility. 

One of the strategies is to follow the industry’s best practices. Let’s look at them now.

Best Practices For Lowering PUE in Your Data Centre

Best practices mean the techniques and methods that have given definite results. And that’s the reason why they are obvious first steps for many facilities when it comes to reducing their PUE. 

Manage the airflow

A good data centre design will help you achieve better airflow. There are several data centre design techniques like hot/cold aisle configuration that separate the hot air from the cold air within the facility. 

The facilities may use containment solutions like blanking panels to effectively separate the airflow. 

One more way to optimise airflow is by using concealed and well-defined cable management. Cables that do not interfere with the airflow make it more effective in cooling the facility.

Lifting the servers and racks off the ground is one more way to optimise airflow. With no obstruction and heat being trapped down below, it can make cooling much more efficient. 

Set the right temperatures

If you set lower temperatures across the facility, it’ll require more power to maintain those temperatures. What you can do to save that power is raise the temperatures according to the ASHRAE(American Society of Heating, Refrigerating, and Air-conditioning Engineers) guidelines. 

Every 1° F rise in temperature can give you 4-5% of power savings according to Energy Star. 

If you do that, your facility will be able to operate just as efficiently as before but in a wider temperature and humidity bands. Alternatively, you can consult a data centre expert like Netcon to help you with cooling and temperature optimisations. 

Precision cooling over comfort cooling

Precision cooling is for data centres. Comfort cooling is for offices and homes. Because they are designed to operate for a specific use case, they can control the temperature and humidity with a lot more precision than regular cooling systems. 

With higher airflow rates that they can sustain 24/7, they are more efficient in removing IT heat from the facility.

If you’re data centre has high-density areas, the normal cooling won’t cut it. You must have precision cooling, that too an in-row or in-rack unit.

Have the right power infrastructure

Power is at the centre-stage of a data centre operations. All our efforts are a constant push towards reducing power. But these will go in vain if we don’t have the right power infrastructure. But right, we mean efficient and reliable. 

Modern UPS systems are highly efficient. They operate at 97% efficiency, sometimes at 99% in ECO mode, compared to somewhere between 88-92% for the older UPS systems. If you were using older UPS systems, modern ones can help you reduce your PUE rating by 0.1 to 0.2 points. 

Modular UPS systems go a step further as they allow you to scale your power needs as your data centre grows without any fuss. They have N+1 or N+N redundancy configurations. They improve the efficiency at partial loads and ease the maintenance. 

One more way to reduce the transmission power losses is by having the direct DC power distribution. This will eliminate the multiple conversion steps, resulting in higher efficiency. 

Remember to consider the total cost of ownership over an extended period of time rather than just the upfront costs. And align these upgrades towards the larger efficiency and sustainability goals. 

Switch to LED lighting

LEDs are the most efficient lights right now. US Department of Energy says that LEDs use upto 90% less energy and last upto 25 times longer than the traditional lighting. 

These modern lights coupled with the occupancy sensors that dynamically manage the lighting can help you lower the PUE rating significantly, especially if you’re switching from traditional lighting to the LEDs. 

And having a design that uses as much daylight as possible will directly save the energy spent on lighting systems. 

Maintain the facility and the systems regularly

What does a maintenance check look like in a data centre? Regular thermal imaging of the facility will reveal the hotspots, areas that aren’t being cooled properly. 

For the cooling systems, replacing the air filters regularly can help boost their cooling power. And just a regular check on all the critical systems, airflows, and cables can help you avoid mishaps. 

Monitor the metrics of your data centre

PUE is not a one-time measuring metric. It constantly changes, as we discussed about it in the previous blog. Factors like weather, performance output, load, etc affect it. 

Regular PUE audits, let’s say weekly or at least monthly, can help you clearly gauge the health and efficiency of the facility. Data Centre Infrastructure Management softwares can be your ally to monitor your data centre and its key metrics. 

AI can also help you maintain your data centre. It can give you predictive analysis after studying the data from all the sensors across the facility. With this data, you can optimise your facility for better PUE numbers. 

Now that we’ve looked at the best practices, let’s focus on how the innovations can impact PUE. 

Technological Innovations and Their Impact on PUE

What’s at the bleeding edge of technology now will immediately overshadowed by something even more revolutionary and innovative. But that doesn’t mean we constantly keep updating our infrastructure. 

However tempting the technology or innovation is, you should only upgrade after calculating its total cost of ownership and how much ROI it’ll generate for you. 

Let’s look at some of these innovations. 

Data centre cooling innovations

Liquid cooling is one such innovation. You can look at it in three ways. 

One is the direct-to-ship liquid cooling. As the name tells you, it’s used to cool the high-performance servers quickly and efficiently so they can keep performing at their highest clock speeds.

When you go for immersion cooling, you ditch the traditional air cooling infrastructure and immerse your server and storage racks in a dielectric cooling liquid. Liquids conduct and dissipate heat more efficiently than air. And this will reduce the PUE rating. 

Two phase immersion cooling is a kind of immersion cooling where the dielectric cooling liquid gets in contact with the hot servers, turns into vapour and then condenses back to its liquid state on a heat exchanger. 

Vapour cooling is one more way of cooling a data centre facility, especially if it’s located in a dry hot region. It usually uses water as a wet medium. The air passed directly or indirectly through this medium and cools the facility. 

If it’s direct vapor cooling, it’ll add humidity which is good for a dry arid region. If it’s indirect vapour cooling via heat exchangers, it won’t add humidity and is slightly less efficient than direct vapour cooling. 

Geothermal heat exchange cooling systems work in a similar fashion but they use the earth’s constant temperature, usually at a depth of 20+ feet. One benefit of this is that it can be used for both heating or cooling, depending on the ambient temperature. 

All these innovative cooling methods function better over the traditional air cooling method. 

AI and machine learning

Everything in a data centre mostly comes back to how efficiently you can cool your facility. 

And AI and ML algorithms can help you achieve that last few percentages of efficiency. They can dynamically optimise the cooling with changing loads. They can even predict loads times and prepare your facility to cope with it. 

Let’s say your facility experiences the maximum loads between 3 PM and 5 PM for the majority of the days. AI algorithms will learn this, automatically balance the load and optimise the cooling to provide the maximum performance at the best possible efficiency. 

What’s great is that they can also predict potential issues or maintenance needs, reducing your downtime.

Power management innovations

Using software-defined power management to dynamically allocate power will lead to lower power consumptions for your facility. 

Lithium-ion batteries with densely packed cells will make your UPS and power storage more efficient. These innovations are worth considering if you’re struggling to achieve better PUE numbers. 

Computing innovations

Processors are getting more and more dense when it comes to performance. And they are becoming more energy efficient as well. 

If you compare these modern processors, with their 3 nanometer process, with the old ones, they can provide much more performance than the old processors while using the same or less amount of power. Processing power per watt has gone up significantly. 

Heterogenous chips with dedicated processors for different tasks are also becoming common. ARM architecture is becoming mainstream. The best examples are Apple’s M-series of chips. 

You can certainly take advantage of these innovations to bring your power consumption down, and eventually achieve a lower PUE rating. 

Now that we know the ways to lower the PUE rating of a data centre, we learn from the best around the world on how they do it. 

Case Studies on PUE Improvement

You may have a PUE rating of 1.5. That rating may be fairly consistent across the year. But that doesn’t mean that’s the best you can do.

These case studies help us learn what the best data centers around the world do to keep themselves as efficient as possible. 

How Leading Data Centres Achieve Low PUE

Google's data centers are setting the bar high with a trailing twelve-month PUE of 1.10 across their facilities. 

They're not just implementing best practices; they're rewriting the rulebook. Google custom-designs their servers and power supplies, optimizing every component for maximum efficiency. But what sets them apart is their use of advanced machine learning for cooling optimization. 

This AI-driven approach allows them to fine-tune their cooling systems in real-time, adapting to changing loads and environmental conditions.

Facebook (now Meta) is another tech giant pushing the PUE envelope, reporting figures as low as 1.08 in some facilities. Their secret weapon? The Open Compute Project. 

This initiative for efficient hardware design goes beyond just Facebook, fostering industry-wide collaboration to drive innovation. They're also leveraging their locations smartly, using ambient air cooling in cold climates to minimize the need for energy-intensive cooling systems.

Microsoft is thinking outside the box – or rather, inside the ocean. Their underwater data center project, Project Natick, is a bold move towards ultra-efficient cooling. 

By submerging servers in the cold depths, they're eliminating the need for traditional cooling methods altogether. On dry land, they're pioneering the use of hydrogen fuel cells for backup power, a cleaner alternative to diesel generators that could reshape how we think about data center resilience.

Amazon Web Services (AWS) takes a different approach. While they don't publicly disclose PUE figures, their focus on overall efficiency is clear. 

They're tailoring their cooling systems to different climates, recognizing that a one-size-fits-all approach won't cut it when you're operating globally. In water-stressed regions, they're using recycled water for cooling, showing that efficiency and sustainability can go hand in hand.

Lessons from the Most Efficient Data Centers in the World

Supermicro's Silicon Valley data center is a masterclass in efficiency, boasting a PUE of 1.06. Their success highlights a crucial point often overlooked: the importance of holistic design optimization. 

It's not just about having the latest cooling tech. It's about how all the pieces fit together. The close collaboration between IT and facilities teams at Supermicro has been key to their success, breaking down silos that often hinder efficiency efforts in larger organizations.

Green Mountain's DC3-Oslo facility takes the crown with an astonishing PUE of 1.02. The secret to their success? Location, location, location. 

Norway's cold climate and abundant hydropower allowed them to create a naturally efficient environment for data storage. But it's not just about picking the right spot. Their use of seawater for cooling showcases how innovation can push efficiency even further in ideal conditions.

Yandex's data center in Finland, with a PUE of 1.07, has combined direct free air cooling with adiabatic systems, a smart move in a climate that varies significantly throughout the year. But what's particularly a highlight is their investment in custom server design. This approach, while requiring significant upfront investment, pays dividends in long-term efficiency.

Switch's Las Vegas data centers prove that impressive efficiency is possible even in challenging environments. Despite the scorching desert heat, they've achieved a PUE of 1.28. 

Their success lies in proprietary cooling technology and a commitment to continuous innovation. It's a reminder that with the right approach, significant efficiency gains are possible even in the most unlikely places.

These case studies reveal some common threads in achieving exceptional PUE ratings:

1. Context is king. What works in Norway might not work in Las Vegas. The most efficient data centers tailor their approaches to their specific circumstances.

2. Innovation is non-negotiable. Whether it's underwater servers or AI-driven cooling, pushing the boundaries of technology is key to pushing down PUE.

3. Holistic thinking yields results. The best performers don't just optimize individual components; they consider how all aspects of the data center work together.

4. Collaboration drives progress. Whether it's internal teamwork or industry-wide initiatives like the Open Compute Project, sharing knowledge accelerates innovation.

5. Efficiency is an ongoing process. Achieving a low PUE isn't a one-time effort. It requires continuous monitoring, optimization, and investment in new technologies.

As we look to the future, it's clear that the race for efficiency is far from over. But it's also evident that the focus is shifting. 

While PUE remains an important metric, leading data centers are increasingly considering broader measures of sustainability, including water usage, carbon footprint, and overall environmental impact. 

The next frontier in data center efficiency may well be balancing these competing priorities to create truly sustainable digital infrastructure.

Calculating and Monitoring PUE

Now that we've explored how industry leaders are achieving impressive PUE ratings, let's dive into the nuts and bolts of calculating and monitoring PUE. This knowledge is crucial for any data center operator looking to improve their efficiency.

Step-by-Step Guide to Calculating PUE

Calculating PUE might seem straightforward, but there's more to it than meets the eye. Let's break it down:

1. Understand the formula: PUE = Total Facility Energy / IT Equipment Energy. Simple, right? But the devil's in the details.

2. Define your boundaries: Before you start measuring, you need to know exactly what constitutes your data center facility. Where does the utility handoff point lie? This seemingly small detail can significantly impact your calculations.

3. Identify all energy sources: It's not just about the electricity bill. Are you generating power on-site with solar panels? Using natural gas for heating? Account for everything.

4. Measure IT equipment energy: This isn't just servers. Include storage, network equipment, and even those often-forgotten monitoring workstations.

5. Choose your measurement interval: While annual PUE is the gold standard for reporting, don't neglect shorter intervals. Daily or weekly measurements can reveal seasonal trends or sudden efficiency drops.

6. Collect accurate data: Invest in good energy meters. Estimates might be tempting, but they can lead you astray.

7. Crunch the numbers: Apply the formula consistently. Stick to one unit of measurement (typically kWh) to avoid confusion.

8. Interpret your results: As of 2022, the industry average PUE was around 1.57. If you're hitting 1.2 or lower, you're in excellent territory. But remember, context matters. A PUE of 1.3 in a hot climate might be more impressive than 1.2 in a naturally cool location.

9. Document everything: Record your assumptions, methodologies, and any exclusions. Transparency is key, especially if you're reporting your PUE publicly.

Here's an expert tip: Consider calculating partial PUE (pPUE) for specific systems or areas within your data center. This can help you pinpoint inefficiencies more precisely.

Already familiar with calculating PUE? Here's a blog on how you can master PUE for better data center management.

Tools and Technologies for Monitoring PUE

Calculating PUE is one thing; monitoring it consistently is another ball game entirely. Let's look at some tools that can help.

Data Center Infrastructure Management (DCIM) Software 

Think of DCIM as the Swiss Army knife of data center management. Products like Schneider Electric's EcoStruxure IT or Sunbird's dcTrack offer real-time PUE monitoring, historical data analysis, and even predictive analytics for optimization. 

But here's what many blogs won't tell you: the real power of DCIM lies in its integration capabilities. When connected to your building management systems, it can provide a holistic view of your entire operation.

Power Monitoring Systems 

High-precision power measurements are crucial for accurate PUE calculations. Systems like ABB's Ability Electrical Distribution Control System don't just measure power; they can analyze power quality and create detailed load profiles. This granular data can reveal inefficiencies that broader measurements might miss.

Environmental Monitoring Systems 

Products like APC's NetBotz go beyond simple temperature and humidity monitoring. They can detect air flow anomalies and even water leaks. When integrated with your cooling systems, they can help optimize your cooling efficiency in real-time.

IoT Sensors and Smart PDUs

Don't underestimate the power of smart power distribution units (PDUs). Raritan's Intelligent PDUs, for instance, can provide rack-level power monitoring. This granularity can be a game-changer, especially in identifying energy-hogging equipment or imbalanced loads.

Artificial Intelligence and Machine Learning Platforms 

This is where things get really interesting. Platforms like Google's DeepMind AI aren't just monitoring; they're actively optimizing. They can predict maintenance needs, dynamically adjust cooling based on workload, and even detect anomalies in power usage patterns before they become problems.

Here's a pro tip that often gets overlooked. 

While these advanced tools are powerful, they're not plug-and-play solutions. The real value comes from having skilled professionals who can interpret the data and make informed decisions. The best data centers have teams dedicated to analyzing this information and continuously tweaking operations for maximum efficiency.

Another point to consider: while PUE is important, it's not the only metric that matters. Forward-thinking data centers are also monitoring Water Usage Effectiveness (WUE) and Carbon Usage Effectiveness (CUE). These give a more comprehensive picture of a facility's environmental impact.

As you implement these tools, keep scalability in mind. Your monitoring solution should be able to grow with your data center and adapt to new technologies. And don't forget about data security – your monitoring tools will have access to sensitive operational data, so robust security features are a must.

Lastly, remember that monitoring is an ongoing process. Regularly calibrate and validate your tools to ensure accuracy over time. The most efficient data centers aren't just monitoring PUE; they're constantly looking for new ways to optimize it.

These tools and technologies, along with the insights they provide, can help you join the ranks of the world's most efficient data centers. It's a journey of continuous improvement, but the rewards – both financial and environmental – are well worth the effort.

How Netcon Technologies Can Help

Optimizing your data center's PUE is not a one-time effort, but a continuous journey of improvement and innovation. As we've explored in this guide, there are numerous strategies and technologies at your disposal to enhance your facility's efficiency.

From implementing best practices in airflow management and cooling to leveraging cutting-edge innovations like AI-driven optimization and liquid cooling, the path to a lower PUE is clear. But it requires dedication, expertise, and a willingness to embrace change.

Remember, every data center is unique, and what works for one facility may not be the optimal solution for another. That's why it's crucial to have a partner who understands the nuances of data center efficiency and can tailor solutions to your specific needs.

Our team of experts is here to help you navigate this complex landscape. We offer:

• Comprehensive PUE assessments to identify areas for improvement

• Customized efficiency strategies tailored to your facility's unique characteristics

• Implementation of advanced cooling and power management solutions

• Ongoing monitoring and optimization services to ensure continued efficiency gains

Don't let the complexity of PUE optimization hold you back from achieving world-class efficiency. With the right approach and expert guidance, you can transform your data center into a model of efficiency and sustainability.

Ready to start your journey towards optimal PUE? Contact us today to schedule a consultation and discover how we can help you unlock the full potential of your data center's efficiency. Let's work together to create a more efficient, sustainable, and cost-effective data center operation.