What happened to Microsoft's underwater data center that sank to the bottom of the sea for 2 years?

赵玉田

What happened to Microsoft's underwater data center that sank to the bottom of the sea for 2 years? [Copy link]

Microsoft first proposed the concept of underwater data centers in 2014 , believing at the time that the concept had the potential to provide high-speed cloud services to coastal populations and save energy.

In 2015 , Microsoft demonstrated the feasibility of the underwater data center concept during a 105-day deployment in the Pacific Ocean.

In 2018, Microsoft sank a data center to the icy bottom of the ocean north of Scotland.

In September 2020, Microsoft retrieved a data center from the seabed after two years of trial operation. Researchers evaluated it and the first conclusion they came to was that the server failure rate in underwater data centers is lower than that in traditional data centers.

The whole process is briefly described:

Before entering the water:

Two years later

The Northern Isles data center was lifted from the sea floor near the Orkney Islands in Scotland using a gantry crane:

You can see the surface changes at a close distance:

Towed to a pier in Stromness, Orkney:

A layer of algae, barnacles and anemones grew outside the data center:

Anemones growing in hidden corners of the ballast-filled base:

The strong cleaning of the "Northern Islands" underwater data center has begun:

The data center was filled with dry nitrogen before being sealed, separating it from its ballast-filled base, ready for transport to land:

Removing the pipe end caps from the underwater data center:

Looking inside an underwater data center:

A server was removed from an underwater data center and will be analyzed to help researchers figure out why servers in underwater data centers are eight times more reliable than servers in the same data center on land.

Beginning the removal of 12 server racks and associated cooling system infrastructure:

Microsoft said that its current research focus is on how to solve the problems of metal cabin corrosion and biological fouling. After all, if metal is left in seawater for a long time, it will slowly corrode and affect data security.

The first, of course, is keeping the inside of the data center's large steel containers dry. The second is figuring out the best way to use the surrounding seawater to cool the servers inside. And finally, there's the question of how to deal with the barnacles and other marine growth that inevitably clings to the underwater containers.

At present, Ben Cutler said that underwater data centers have passed the stage of scientific experiments, but there is still a question worth pondering: should we design a large underwater data center or a small underwater data center?

In addition, Microsoft said it will be cautious about when underwater data centers can be commercialized, and is confident that it can prove the ultimate value of this idea.

I hope that there will be more of these seemingly absurd but actually effective data storage methods. After all, with the rapid development of society today, dataization is an inevitable trend. More and more data means more and more load. Only when there are more stable and long-term storage methods can there be better solutions for data maintenance and retention, which is also an important step for the inheritance of human civilization in the future.

[Source: Crazy Mechanical Control]

qwqwqw2088

How much lower can the failure rate of such a labor-intensive data center be than that of a traditional data center?

吾妻思萌

I think it's great to have the courage to try. I used to think it's impossible to fly to the sky? Now I'm flying on a plane.

Honghuzaitian

qwqwqw2088 posted on 2022-11-8 08:50 How much lower can the failure rate of such a laborious data center be than that of a traditional data center

It should be mainly to save energy. Traditional data centers consume a lot of energy for heat dissipation. This direct use of natural water cooling may not be noticeable in the short term, but it can save considerable economic expenditure in the long run.

qwqwqw2088

Honghuzaitian Published on 2022-11-8 09:22 It should be mainly to save energy. Traditional data centers consume a lot of energy on heat dissipation. This kind of direct use of natural water cooling will not be visible in a short time, but it can be seen in the long run...

Mainly to reduce heat dissipation, it is not easy to repair

BAT has all its servers in Guizhou. The average temperature in Guizhou is around 25℃ in summer and close to 10℃ in winter.

eew_1PrAA0

The failure rate is low when it is put underwater, but it is inconvenient to repair it outside. There is still plenty of space in the ocean.

UUC

This should be the future development direction of data centers. After all, it has a great advantage in energy saving. As for maintenance and failure, there should be a big breakthrough as the electronics industry develops.

秦天qintian0303

The question is how to deal with the interface? It is OK to seal it directly, but it is difficult to deal with the interface.

Honghuzaitian

qwqwqw2088 posted on 2022-11-8 09:27 The main thing is to reduce the heat dissipation, which is not convenient for maintenance. BAT is all in Guizhou. The average temperature in Guizhou in summer is about 25℃, and in winter...

It is not very convenient to repair. If the stability is good and there is no problem with the hardware, this solution is still feasible.

476700838

Microsoft's official response is to save costs, the most important thing is to save electricity

数码小叶

Time flies so fast, it seems like it was just yesterday that we went into the water