It took a whole day for Microsoft to heave one of its data centres, roughly the size of a large shipping container, 117 feet up from the bottom of the North Sea.
Covered in algae, clusters of barnacles and sea anemones, the “data warehouse” had been sitting off the coast of the Orkney Islands since spring 2018 – all part of an experiment conducted by Microsoft’s Project Natick research team to see whether it was possible, operationally and logistically, for a data centre to work underwater.
The crux of the experiment came down to sustainability. As one of the world’s fastest-growing real estate asset classes, investment in data centres reached £23.5bn in the first six months of this year, according to data from Knight Frank, already superseding last year’s total investment volume, which stood at £1.5bn. And according to financial services firm M Capital Group, the data centre industry is expected to grow by 10-15% over the next five years.
But there is a problem: given the nature of their function and high levels of energy usage, these facilities are well known for being extremely unsustainable. The biggest may require more than 100 megawatts of power, which is enough to power 80,000 homes. McKinsey & Company research shows that data centres have been known to emit a massive 80 megatonnes of carbon dioxide a year.
Early indications show that, as the sea naturally maintains a cool temperature, data centres could need roughly 20% less energy to cool the servers down than their on-land counterparts.
The Project Natick experiment is Microsoft’s shot at a solution. Under the research tagline of “50% of us live near the coast. Why doesn’t our data?”, the research team is focused on testing the feasibility of subsea data centres powered by offshore renewable energy.
The retrieval of its submerged centre this year marked a significant step forward. If the experiment proved successful, it could potentially shake up the future of the entire data centre market.
Under the sea
After cleaning the marine life from the data centre, the Project Natick team discovered that not only did the data centre survive underwater, it actually operated much more reliably than data centres on land.
The centre’s servers were eight times more reliable compared with those on land, according to Ben Cutler, a project manager in Microsoft’s Special Projects research group.
Corrosion from oxygen and humidity, temperature fluctuations and “bumps and jostles from people who replace broken components” contribute to equipment failure of data centres on land, but the underwater data centre, which was pumped full with dry nitrogen and maintained at a cool temperature, managed to avoid these variables. “It is a pretty compelling vision of the future,” said Cutler.
It will take more than one two-year experiment to know for sure just how feasible this sort of solution is on a wider scale, though. So how likely is it that data centre development could see a large-scale, global shift from land to sea?
It will be an uphill struggle. People don’t want to deal with an initiative that is too radical
Jon Summers, Research Institutes of Sweden
Edging towards deep data
For some experts, including Andrew Jay, head of EMEA data centre solutions, advisory and transaction services at CBRE, the prospect of putting data into the deep is not as far-fetched as it sounds; but there are limits.
Hyperscale data centres – those that run on 100 megawatts of power or more – “can’t possibly” go underwater, he says, adding that it would be unlikely that colocation data centres (those close to capital cities) could be submerged as they often have people working in them and need to be within 35 miles of the core market they serve. But he says that Microsoft’s experiment could be a game-changer when it comes to deciding where to locate edge data centres – those that serve localised areas.
Edge data centres, he adds, are an emerging asset class within the sector. This is because the general population is using more and more data, a trend that is only set to accelerate as a consequence of Covid-19.
As it becomes “more difficult to pump these big files around, all over the world”, Jay says there is a need for smaller data centres to serve local areas outside of the major cities. “This is where the underwater data centre could be really exciting,” he says. But would the effort of plunging them into the sea be worth the benefits? Jay thinks so, especially as it addresses some of the key sustainability concerns surrounding the asset class.
The amount of energy that data centres use to operate is measured in terms of power usage effectiveness. Colocation data centres, according to Jay, usually run at a PUE of 1.2, which means that data centres need an extra 20% of electricity to cool their servers down.
To make data centres as sustainable as possible, providers need to get their PUEs as close to 1 as they can. Jay predicts that Microsoft’s underwater data centre has been successful in lowering its PUE, because less energy is needed to reduce the temperature of servers in the cool and stable temperature of the sea.
Microsoft pledged in January last year that it will take “aggressive approaches” to be carbon negative by 2030 and announced in July this year that it was “completely changing” the way it operates its data centres.
An uphill struggle
There are other benefits in creating underwater edge data centres, according to Research Institutes of Sweden’s scientific lead and data centre expert Jon Summers. He points out that it will be a lot harder for people to interfere with a tightly sealed underwater data centre compared with one above ground.
It could have economic benefits too. With an anticipated boom in data centres as consumers use more and more data, providers will be on the hunt for cost savings. Placing a data centre underwater for a prolonged period of time, says Summers, could reduce maintenance costs significantly.
However, for every benefit there is a cost. “It will be an uphill struggle. People don’t want to deal with an initiative that is too radical. These are multi-billion-dollar ventures, and you want to use tried-and-tested technology that you know will deliver,” says Summers.
This commercial viability is called even more into question considering that there are less time- and resource-intensive ways of cooling down and maintaining data centre servers than plunging them underwater and pumping them full of nitrogen. He says technology is already in development to tackle this, such as liquid cooling techniques.
Like a man on the moon
Garry Connolly, founder of Host in Ireland – a body that looks to increase digital investment in the country – is even more sceptical about whether the future of data centres lies beneath the waves.
His main concern surrounds the risk of failure. “Data is currently the oxygen of the global economy,” he says. “You throw a data centre into the Hudson River, the Liffey, the Thames, where you have active ports, ships and boats – it’s a whole different dynamic of risk.”
Instead, he likens the experiment to the moon landing, in that it will be more impactful in helping improve on-land facilities than it will be in a quest to put them underwater. Putting man on the moon wasn’t a quest to try and establish a new civilisation in space, Connolly says. It was predominantly done for the scientific learnings scientists would gain from the experiment, from understanding more about how gravity works to discovering that people lose body mass in space. Connolly thinks the same stands for Microsoft’s experiment. “It gave them great findings, analysis and information for the production of [on-land] data centres,” he says.
Instead of investing in the development of underwater data centres, Connolly suggests money could be better spent on more groundbreaking and innovative research programmes to transform the sector. “The ultimate Holy Grail, and something which isn’t as far-fetched and scientifically Frankenstein-y as you might think, is the storage [of data] on synthetic DNA,” he says.
Looking to experts on research in this field, the savings on size could be massive: the chief executive of SynBioBeta (a community of investors, scientists and engineers involved in synthetic biology) penned an article in Forbes reporting that just one gram of synthetic DNA can store 215m gigabytes of data.
Summers also agrees that technologists and scientists have bigger issues to solve in the sector before embarking on a mission to put facilities into water. “I think the real crux of the problem is from an engineering perspective,” he says. “We talk about reusing heat and better cooling systems, but why are we creating so much heat in the beginning? That’s where the real engineering needs to be applied.”
So, although Microsoft has proved it is perfectly possible to plunge data centres underwater, the question as to whether we should remains unanswered.
From a sustainability perspective it makes perfect sense, security breaches may well be reduced and there are some potential economic gains. But whether the real estate sector is prepared to take on the additional cost of something that many might argue is not yet commercially viable could well prove to be a fly in the ointment, at least for the foreseeable future.
To send feedback, e-mail lucy.alderson@egi.co.uk or tweet @LucyAJourno or @estatesgazette
