Executive Summary
- For most of us, an unexpected turn in the forecast is a minor inconvenience.
- For a data centre, a bad forecast can mean rather more than getting a bit damp on the school run.
- As we head into the autumn/winter months, these examples show that a weather problem can take many forms, and how we can be ready for it…
We Brits love talking about the weather. We complain when it’s too hot, complain when it’s too cold, and, when it comes to rain, well, we’re hardly short of things to complain about.
We’ve even managed to turn the weather into a national excuse. There was the infamous “wrong type of snow”, of course, which became shorthand for British infrastructure being caught out by something that, depending on your perspective, was either entirely predictable or absolutely unprecedented…
Sometimes it directly affects a data centre, sometimes it creates a much bigger problem somewhere in the surrounding infrastructure, and sometimes the failure barely registers inside the data centre at all because the backup systems have done exactly what they were supposed to do.
After all, we may joke about the wrong type of snow, but for an industry increasingly responsible for keeping everything from our bank accounts to our hospitals running, we need to make sure we have the right type of infrastructure to deal with whatever the weather decides to throw at us.
When the River Aire came calling
Let’s start in Leeds, where December 2015 brought Storm Eva, the bursting of the River Aire’s banks and some of the worst flooding the city had experienced in years.
Among the buildings caught up in the flooding was Vodafone’s data centre on Kirkstall Road. The site lost power and had to rely on emergency arrangements, but the flooding created another, rather less obvious problem: getting the people needed to restore those services to the building was suddenly rather difficult.
Engineers were working from Boxing Day onwards, while recovery equipment, including generators, had to be brought in once access to the site was possible. The incident affected voice and data services for some customers, while the wider flooding also disrupted communications infrastructure used by emergency services and other organisations.
When we talk about data centre resilience, we tend to picture the building itself, but what about the roads leading to it? If that road is underwater, it doesn’t matter how many engineers you have on standby if they can’t actually get to the equipment that needs fixing.
During the same period of severe flooding, a BT exchange was also overcome, affecting tens of thousands of fixed-line and broadband customers. Emergency communications were disrupted and engineers worked through the Christmas period to restore services.
Data centres are not as self-contained as we think. Inside are the servers, outside are the generators, and somewhere nearby are the engineers who know how to keep the whole thing running.
The flooding didn’t need to reach every individual piece of equipment to cause disruption. It simply needed to affect enough of the infrastructure around it.
Storm Arwen and the very long power cut
Storm Arwen arrived in November 2021 with winds strong enough to cause widespread damage to Britain’s electricity distribution network, leaving more than a million customers without power and tens of thousands waiting several days for their supplies to be restored.
Backup generation exists precisely because the grid cannot always be relied upon, but a generator is only useful for as long as it has something to burn, and getting fuel to a site can become considerably more complicated when roads are blocked, engineers are dealing with hundreds of other incidents and large parts of the surrounding infrastructure are struggling too.
It’s one thing to have a backup generator ready to go when the power goes off. It’s another to consider what happens if the power stays off for considerably longer than expected, or if getting fuel, equipment or people to the site becomes difficult.
So, when we talk about data centre resilience, are we preparing for the power going off, or are we preparing for the possibility that the power could stay off for considerably longer than we’d like?
The time the grid failed – and the data centres didn’t
In March 2025, a major fire at the North Hyde electrical substation in west London caused a significant power disruption. Heathrow Airport closed, homes and businesses lost power and Hillingdon Hospital was affected, while the wider transport network had to deal with the consequences of the outage.
Three data centres also lost their normal grid supply, but they kept running; switching to backup generators and continuing to operate while the grid supply was unavailable.
Here, the most successful response to a major infrastructure failure is the one that hardly anyone notices.
The grid goes down, the backup systems kick in, and the services people rely on continue running.
Resilience isn’t simply about preventing every possible failure, because that would be an impossible task. Sometimes it is about accepting that something outside your control is going to fail and making sure that failure doesn’t become your problem.
It’s not always what happens inside the data centre
A data centre can have backup power, resilient cooling, multiple layers of protection and all the other systems designed to keep it operational, but it still relies on a much bigger network around it.
It needs electricity from somewhere. It needs telecommunications infrastructure, fuel, engineers who can get to the site, and roads that allow them to get there.
This doesn’t mean that every storm is going to turn into a major incident, or that every data centre is one flooded road away from disaster.
But it does mean that resilience planning has to look beyond the walls of the building.
