In The News / ITN - Each week I pick a new story from the news and provide my thoughts and opinions on it.
Yes that’s right, Microsoft are putting one of their datacenter’s albeit a small one under the sea. The shipping container sized pod contains 864 servers and is designed to run uninterrupted for 5 years.
Is It Just A Stunt?
Not entirely, although they could do the same task without inviting the press, so its obvious they’re wanting to show the project off but why wouldn't you, it so cool!
As bazaar looking as it is, it’s a very smart idea. Datacenter's use huge amounts of power for both running servers and cooling them down. 3% of all power consumed in 2016 globally was by datacenters and this figure is only going to increase. This means the race is on to create more power efficient methods and use sustainable power sources to operate them.
One option which Google and Facebook are using, is to locate their datacenter’s in colder climates such as Finland and Sweden. This works out great for saving on your cooling bill, but you can’t locate all you centres in cold climates or people near the equator would experience horrendous latency.
Having sites based at sea allows Microsoft to use the constant cold flow of water to provide the necessary cooling, whilst letting them be located at key geographic regions around the world to best service their customers.
Fully Renewable?
Ok so they’ve taken care of the cooling by using the sea, but how are they actually going to power all those servers?
The Orkney Islands in Scotland, is where the trial is taking place. It’s also no coincident that the Orkney Islands are a global research location for sea based renewable energy. Since the 1950’s the people of Orkney have been trying to take advantage of their location and weather to generate power.
The ultimate goal here is to combine the pod with some form of wind / tidal or solar based power generation. By achieving this they would create the fully sustainable datacenter which would not be reliant on being located near land based energy sources.
What Could Go Wrong?
I know what you’re thinking, ‘This all sound perfect on paper but it is actually feasible in the real world?’. Like anything ambitious there are obstacles to overcome.
Reliability
The most significant of those obstacles is reliability. If you’ve placed your data centre on the sea bed nowhere near land and one of your servers decides to break, you cant just go and turn it off and on again. Microsoft says it is designing the pods to not require maintenance for 5 years. This is a very long time especially for high performance parts, so I hope they have a rigorous testing process.
It would also be very interesting to know how they plan to future proof the design to stay suitable for the computing demand required in 5 years time.
Security
Whilst land based datacenters are easier to find, they are still very hard to access due to the multitude of internal and external security measures. However, your data centres are on the sea floor it defiantly stops ‘convenience’ thief's and unauthorised physical access. Breaching the pod would just destroy the entire thing instantly due to the pressure and that water & computers don't mix. All this would happen before you could do anything malicious.
One threat it would be open to is that of international espionage. There has been a sharp rise in the reports of countries taking a keen and close look at the undersea cables which provide 97% of global communications. By placing large amounts of compute power down there, not just the transmission cables you are potentially putting more at risk.
Damage
Just because something is on the sea bed doesn't mean it is out of harms way. Roughly 70% of all damage caused to undersea cables is by fishing activities and ships. This threat would be increased for any sea based datacenter because not only does it require cables connecting to it at each end but its significant bigger than a cable so is a much bigger target for anchors and nets to hit.
Scale
Normal datacenters can house up to 80,000 servers each, so for Microsoft to make this truly viable, they are going to need to find a sufficient method to scale the pod’s current capacity. This could take the form of a much larger single pod or a collection of pods linked together.
Final Thoughts
Anything to further the development of green technology and reduce our energy impact gets a thumbs up from me, throw in that it could provide a better experience for customer too and you really do have a winner. Sure there are problems to work out but isn't there always?
I think this is a great project and one which Microsoft should be commended for pursuing. It’s really a case of thinking outside the box on this one.
You can find more about the project here.
Jack
