10 December 2024

Designing a data centre that can accommodate future sustainability regulations and accommodate next-generation technologies is no mean feat…

Data centre design has come on in leaps and bounds in recent years, with new technologies changing the face of the industry forever.

“Considering the long lifespan of data centres, which can exceed 20 years, and the rapid refresh cycle of IT infrastructure - typically every three years - it is crucial to design with foresight,” asserts Stijn de Kruijf, data centre sustainability expert, Royal HaskoningDHV. “The emergence of AI chip technology, which is already reshaping data centre densities, highlights the necessity for a smart, modular, and flexible design, ensuring the data centre remains relevant and valuable well into the future, potentially up to 2038 and beyond.”

The importance of agility cannot be understated. Jason Koffler, CEO & Founder, Critical Power Supplies, reports that the pace of technological change means infrastructure must support rapid scalability without compromising reliability or security.

“Flexibility in design — whether through modular elements, adaptable power solutions, or flexible cooling systems — allows data centres to evolve alongside emerging tech demands and deliver consistently high performance to end-users. A future-ready design is, at its core, an adaptable one,” notes Koffler.

“Being future-ready demands adaptability, particularly ensuring that data centre operations are fully optimised and run efficiently in a risk-free manner,” agrees Stu Redshaw, Chief Technology & Innovation Officer, EkkoSense. “We need to recognise that, as our industry undergoes significant change, we’re not in a steady state anymore!”

Stu Redshaw - EkkoSense

What’s missing?
Amongst the infrastructure, regulations, power, connectivity, and the challenges of incorporating next-gen technologies, some aspects of data centre design – like redundancy and resilience - don’t get the attention they deserve.

“Resilience is often the underappreciated element of data centre design, particularly in power and cooling infrastructure,” opines Koffler. “As operational demands increase, these systems need a robustness that can endure both peak loads and the unexpected. Redundancy, battery backup, and proactive maintenance are essential to avoiding costly downtime. Investing in scalable power systems from the outset can save millions in the long run, while ensuring a consistently high standard of service.”

Smart building systems, too, are oft overlooked in data centre design. According to Andy Putnam, Services Director (Data Centres), SKANSA, however, this is changing.
“These systems can have a huge impact on increasing the operational and energy efficiency of a facility. Designing and building them in from the start gives clients access to live operational information from day one,” shares Putnam. “This means they can operate as efficiently as possible, including configuring a digital twin to enhance the operational efficiencies in both white and grey space. Meanwhile, having the ability to take this to a central headend creates the backdrop for the use of AI to maximise potential of all the systems and look for energy saving opportunities across multiple sites.”

“What concerns me is that so many data centres still rely on building management system (BMS) levels of sensing to monitor the environmental performance of what now are clearly key strategic assets,” says Redshaw. “Less than 1% of data centres have gone to rack-level telemetry, indeed some rooms were originally specified just to have tens of sensors. This introduces levels of generalisation in terms of telemetry that simply can’t keep pace with evolving IT requirements. We’re now operating in a space where we should be relying on hundreds if not thousands of sensors per room – otherwise you’re exposing yourself to significant risk.”

“Looking further ahead, the automation of data centre operations, combined with autonomous energy efficiency control optimisation, is expected to become more prevalent,” adds de Kruijf. “These advancements enable a high-security regime with minimal personnel presence near valuable data, ensuring 24/7/365 operational focus and uptime. Developing a strategic vision on these topics and translating them into data centre design principles is crucial but not yet widely adopted.”

Accommodating AI
“We expect the rise in AI to be the biggest single factor to influence the design of future-ready data centres,” warns Putnam.

AI demands higher processing power and a level of flexibility in power distribution and cooling that is much greater than conventional systems.

“AI processing could result in a seven or eight-fold increase on current levels of loads,” explains Putnam. “We used to discuss rack loads of 3-4kWatts, this has now increased to 20-30kWatts per rack, and even higher with high-performance computing (HPC). Designing for AI significantly changes both the power and cooling requirements of a data centre. So, they will need to be designed to potentially accommodate much greater power capacity while also incorporating the most effective, and efficient, cooling systems, including DLC and immersion cooling.”

“Use advanced cooling technologies and energy-efficient hardware,” recommends Rajesh Sennik, a partner at KPMG UK. “Also, think about how excess heat can be recycled for other means; for example, can it be used for heating a building?”

Notably, de Kruijf highlights that one of the most pressing issues is the sheer weight of AI hardware: “liquid cooling systems, essential for managing the intense heat generated by AI workloads, can significantly increase the weight of individual racks. Traditional data centre designs, often based on standardised wardrobe-style racks, may struggle to support these heavier loads, leading to structural concerns and potential risks.”

Rajesh Sennik - KPMG UK

Standards driven innovation
Data centre standards are more important than ever before, particularly with larger and medium-sized data centres now having to collect and report on all the data necessary to be compliant with increasing ESG regulations.

“Regulatory and operational standards play a foundational role in data centre design by shaping best practices and ensuring resilience, safety, and security,” comments Koffler. “Standards like ISO 27001 for information security or ISO 14001 for environmental management are essential, not just as compliance measures, but as benchmarks that enhance operational trust and efficiency.”

“Recent regulations, such as the Building Safety Act 2022 and the Higher-Risk Building Regulations 2023 in the UK, have set foundational standards for data centre design,” adds de Kruijf. “While these regulations have not drastically altered design practices, they underscore the importance of safety and risk management.”

However, the evolving regulatory environment, driven by government targets for sustainability and net-zero emissions, is poised to introduce more demanding requirements.

The need for accurate, real-time reporting has become critical, says Redshaw: “this means not only being able to track key metrics like PUE, CUE, WUE and CER, but also automating the process where possible to reduce reporting burdens. Tracking ESG performance is critical for data centres, but we simply can’t afford to resource large teams of specialists to manually analyse out-of-date reporting spreadsheet. That’s why you need an automated reporting framework.”

In September, the UK’s data centres were classified as critical national infrastructure (CNI), joining the emergency services, finance and healthcare systems, and energy and water supplies.

“Greater regulatory oversight may follow, so the role of standards is likely to increase,” warns Putnam.

“Now that data centres are classified as CNI, they are required to be resilient to physical disruptions, such as from extreme weather events or disruptions from potential protests. As such, the design of these facilities will have to be even more secure to prevent any outages,” adds Sennik.

“The CNI status not only recognises the importance of data centres but also imposes a responsibility to adhere to higher standards,” agrees de Kruijf. “It ensures that data centres are resilient, secure, and capable of supporting essential services. However, it also raises questions about what qualifies as CNI. For instance, data centres dedicated to bitcoin mining may not meet the criteria, and this distinction needs careful consideration.”

Jason Koffler - Critical Power Supplies

Keeping it green
New legislation and a global effort to enhance sustainability is placing increased pressure on data centre operators.

“The increasing demand for power which, coupled with the requirements on operators to report on energy use, is driving the need for renewable energy sources, like the use of small modular reactors (SMRs),” reports Putnam. “It’s important that sustainability and carbon reduction strategies are in place through the whole lifecycle - from design, through construction to operational management.”

“Operating the still ‘young’ data centre infrastructure on 100% renewable power, preferably in combination with time matching, is the biggest contribution companies have within their domain,” asserts de Kruijf. “With strategic positioning of energy efficient data centres and blending the data centres into communities (e.g. with heat delivery), data centres can make a difference.”

Koffler believes that a combination of efficient design, renewable energy sourcing, and state-of-the-art cooling technologies are required to make a truly sustainable data centre.

“Leveraging renewable energy sources can dramatically reduce a centre’s carbon footprint, while advanced cooling options — such as free cooling or heat reuse — further reduce environmental impact. Sustainability isn’t just a target; it’s a necessary operational standard that offers long-term savings and aligns with the increasingly eco-conscious expectations of clients and stakeholders,” says Koffler.

Sennik, too, recommends “a structured approach to decarbonising data centre projects. This involves creating a baseline of current carbon emissions, determining a targeted future state, and developing a comprehensive implementation plan to achieve sustainability goals.”

“In addition to these strategies, considering the lifecycle impact of data centre components is crucial. This involves evaluating the environmental footprint of materials from production through to disposal, which includes operational and embodied carbon,” shares de Kruijf.”

One thing’s for sure - “organisations need to get smarter at optimising their data centre performance – there’s nothing worse than an unloaded data centre – it’s like a bus with no people on it!” outlines Redshaw. “Accommodating workload is everything – the more you can bring in, the more you can do, the more efficient you are. This requires true real-time visibility of data centre performance so you know when you’ve brought in as much load as you can safely handle and have optimised it. Getting data centres to a level where they are as efficient as they can practically be is the key to balancing increases in workload demands with sustainability requirements.”