When the Worst Happens: Ensuring Data Centre Resilience
Power outages and grid failures are nothing new, but amid a recent flurry of incidents, there have been rising calls for the data centre industry to be more prepared.
Blackouts that have taken place this year already include the Heathrow Airport power cut and the Iberian Peninsula grid failure – both events highlighting how one single point of failure can result in total disruption.
With such incidents, questions are circulating over how systems can be kept secure, but also how data centres can rely less on the main grid and opt for alternative powers instead.
We hear from experts from some of the leading data centre and engineering companies about the importance of disaster recovery (DR) and how operators can ensure security and stability in the event of a power failure or cyberattack.
A lesson in strategy and resilience
In the data centre, DR refers to a set of procedures and processes to ensure business operations can continue in the event of a disruption or disaster.
During an outage, power disruptions can compromise access control, surveillance, cooling systems and cybersecurity monitoring tools. Therefore, DR strategies and maintaining stable power are often essential in forming the backbone of a company’s continuity plan.
“Every reputable data centre operator will have robust DR and business continuity plans in place – for their services and facilities,” explains Stewart Laing, CEO of Asanti Data Centres. “These plans ensure the continuity of power, cooling, connectivity and environmental controls. But this does not extend to managing DR on behalf of customers, unless explicitly offered as a managed service.
“It’s crucial to clarify where responsibilities lie. Data centre operators are responsible for the resilience of the infrastructure on which their customers rely but customers are responsible for their own DR strategies when it comes to applications, data and business operations.”
Given that data centres often house critical infrastructure for essential industries like healthcare and AI, downtime often has intense repercussions. In this case, DR strategies are vital because they enable what Arturo Di Filippi, Global Offering Director for Large Power at Vertiv, refers to as a continuity of power, in addition to workload migration and resilience.
“While many operators have basic DR frameworks in place, not all are equipped for the speed and intensity of today’s threats, especially in the AI era,” he says. “Data centres must now factor in highly dynamic AI loads, which shift from idle to peak in milliseconds. Traditional infrastructure isn’t always capable of supporting these changes without failure or degradation.
“Operators understand that with AI and high-density computing reshaping load profiles, infrastructure must be both flexible and robust.”
During the Heathrow power outage, data centres local to the sub-station fire continued to provide service as normal.Such a feat speaks to the urgent need for data centres to be designed with self-sufficiency in mind.
Preparation and prioritisation
With DR and cybersecurity becoming increasingly interconnected, as threat actors become more sophisticated each year, the onus isn’t just on the data centre to keep vital information safe.
“Businesses need to ensure they have a strong DR framework that includes frequent backups and clearly defined recovery plans – all of which are crucial in the event of a cyberattack – but it is entirely their responsibility,” Stewart explains. “DR planning acts as a last line of defence against operational disruption, ensuring that services can be restored swiftly and securely if primary systems are compromised.
“This is particularly relevant in today’s threat landscape, where the potential for ransomware, DDoS attacks and data breaches continues to grow.”
Making sure that data centre is ready for a large-scale grid outage is challenging. Particularly in places like the UK, a facility’s primary power source will be the grid until legislation changes to accommodate alternative power sources like microgrids.
As Stewart says, scenario-based planning is critical. “Operators should run simulations, stress-test their infrastructure and ensure that secondary and tertiary systems are ready to take over instantly,” he says. “In addition to hardware resilience, this requires operational agility and well-practised emergency response protocols.
“Ultimately, preparation hinges on the principle of self-sufficiency, in the event of grid failure. The more a facility is prepared for this the better it can protect uptime and meet service-level commitments.”
Arturo adds: “The ability to provide continuous power quality and availability under these conditions isn’t optional. The Heathrow incident demonstrated that high-density infrastructure without sufficient backup leads to rapid systemic failure. Tests must be conducted under realistic conditions, to show the ability to maintain high efficiency and reliability under significant environmental stress.”
With such events in mind, resilience remains mission-critical – both when it comes to physical security and cybersecurity.
“Data centre operators design and build facilities to operate at 100% utilisation. Solutions to do this include a combination or all of the following: on-site power generators, fuel cells and battery energy storage systems (BESS),” says David Watkins, Solutions Director at VIRTUS Data Centres. “But for those technologies to be useful, operators should have appropriate support contracts in place to enable ongoing operations such as refuelling of generators, including diversity of suppliers.
“It is also imperative that operators have the expertise and ability to execute additional maintenance that may be required if generators are required to provide power for extended periods of time. Living quarters on site is also useful to enable staff to stay at the facility, in the event that transportation to and from the data centre becomes challenging.”
Relying less on the grid is necessary, but infrastructure challenges remain
With the recent grid failures in Spain, Portugal and the UK, conversations are now starting to emerge over the need for data centres to reduce their grid dependency, while enhancing power availability.
“The most effective way to reduce grid reliance is to invest in on-site power generation and storage - whether through solar, gas turbines or large-scale uninterruptible power supplies (UPS),” Stewart says. “These systems can significantly reduce operating costs and improve environmental performance. However, current legislation prevents these from being used as a direct power supply.
“The industry must engage with regulators to unlock flexibility in energy use policies. This includes freeing up unused allocations and enabling dynamic power strategies. Until that happens, even the most forward-thinking operators will find themselves constrained by outdated policy frameworks and infrastructure.”
Modern UPS systems are able to support integration with BESS and microgrids so that data centres can be independent from the grid.
“These solutions improve uptime, allow energy flexibility and enable resilience during outages,” Arturo explains. “UPS-centred architectures are now critical to sustaining operations in a world where grid reliability can no longer be assumed. As AI and high-density workloads increase demand and unpredictability, investing in scalable, resilient UPS systems is a proactive step toward long-term infrastructure stability.”
As pressures on the grid intensify, local populations and businesses are also increasing. This can only result in higher demands for power, creating greater unpredictability for those that use it.
“Data centres are already exploring alternative sources of power, including solar, wind or hydroelectric power generation facilities and potentially connection to nuclear power in the form of Small Modular Reactors (SMR’s),” David shares. “There are sustainability benefits to gain from use of alternative power sources, however, most alternative sources of power are not ‘guaranteed’ and are dependent on local climate conditions and the time of day.
“Therefore, it is important to design data centres with a combination of options to ensure continuity of service. The key moving forward will be better grid ‘integration’, as the ability for data centres to move load on and off the grid will be useful for management of the grid when it is under stress.”
Stewart adds: “Progress will require a combination of infrastructure investment, policy reform, and strategic partnerships with energy providers. Only then can data centres fully realise the benefits of a more independent, resilient power model.”
To read the full article in the magazine, click HERE.