15
votes
Microsoft has sunk a data centre in the sea to investigate whether it can boost energy efficiency
Link information
This data is scraped automatically and may be incorrect.
- Title
- Microsoft sinks data centre off Orkney
- Published
- Jun 6 2018
- Word count
- 786 words
I wonder if you could design a completely solid state data center if you knew the temp would be low enough. There's an Isaac Arthur video about colonizing Titan that discusses the amazing efficiency gain from super low temperatures.
No fans and no spinning disks would leave capacitors as the first things to go I think.
I am somewhat worried about heat pollution, especially if this scales up to enterprise use, but if there is a significant reduction in energy consumption, there may be a net benefit regardless.
That being said, I love that it's running on renewable energy, and that the projected failure rate puts the estimated life span at 5 years.
If nothing else, it'll be an interesting find for future archaeologists or treasure hunters!
The data center would produce the same amount of heat above water. From the perspective of global warming I don't think it matters whether the heat is released above the water or below it. And since the ocean is large with currents, I don't think the heating would be very significant outside of a small radius.
Yeah, I don't see it contributing significantly to global ocean temperatures, but a study of the local waters would be nice. Does the minute change in temperature have any effect, basically?
If I had to guess, no. Water has such a high specific heat and is so easily dispersed that it would really surprise me. Plus, with the included efficiency it might actually be better. But then I'm not a marine biologist.
The issue is rising ocean temperatures. The rate at which the ocean is warming is so high and fast, we are losing entire ecosystems due to coral bleaching. If you put the data center in the water, the ocean will absorb more of the heat, faster. It's like putting a pot of water on the stove but turning on a different burner. The temp will eventually rise it's ambient temperature but it would have a much larger impact if its getting direct contact.
I do wonder what happens when one of the servers just breaks or needs to be manually repaired. Are they just going to shut the whole thing down and raise it to the surface? Or are they just going to slowly let them all fail?
I wonder how much Microsoft is paying in power bills, if these folks say they have too much power.
So, if it goes well, they are going to have huge undersea data centres all over the place.
If it goes badly, fish can have a new home off the coast of Orkney.
From what I understand reading it they are going to let them fail. These are probably machines that host virtual machines meaning that failing hardware isn't really as much of an issue as it will just migrate to a different machine.
This is a research project anyway so they don't need an answer for that particular aspect right away.
This is just speculation on my part, but I figure that they probably will do a cost benefit calculation regarding the size of these server containers that makes them big enough to be economically viable even if some hardware fails and at the same time small enough to eventually bring them to the surface with relative ease to replace parts.
I am curious if eventually the cooling benefits will be enough to offset all of the added costs this brings.
This is the stupid "for fun" thing that I'd do if I owned a huge company. If it works it could be good however.
If it sounds stupid and works; it's not stupid.