Project Natick was a successful proof-of-concept, but now it's time for Microsoft to move on.
Corp-speak for a failure. If it was good, you would not stop it (after only 2 years) or not go on with the project in a visible way.
What I guess is that the thing was ok in terms of not being a giant disaster but it did not have any added value compared to a land based équivalent and the external factors costs like maintenance on the server, operating and monitoring the container, ... were terrible compared to a normal datacenter.
Where do you think most of the excess energy from anthropogenic climate change is being absorbed? It's measured on a scale of hundreds of 1e21 J, of which each 1 is 1 million 1 Mt nuclear bombs equivalent. We're already at +200 from 1993. Human activity is leading to massive changes in Earth's delicate systems, and data centers folks want to more than double energy production primarily for AI.
You're not wrong, but not all values of big are equivalent.
e.g. volume of the worlds oceans is ~1e21 L and annual global energy production is ~3e16 Wh = ~1e20 J
1e20/(1e21*4e3) = 0.000025 ΔC
So even all world's energy production dumped into the ocean as waste heat is a minuscule direct effect. It's all in the second order effects of generating that energy...
What about the local effects of raising water temperature? (In the scenario where a large number of underwater data centers were built near each other)
It's generally bad and kills local eco system with algae blooms and messing up seasonal cycles of nutrients/food. But that's just with dumping the heat into environment especially in freshwater river based reactors and a pretty raw system of dumping hot water into the river ...
Clever use of physics and cooling towers allows us to mitigate the problem..release the energy slowly over time or use the thermal for something else.but costs a lot more than just dumping it into ocean and doing lazy misleading napkin math.
Yes of course, local impacts would be more pronounced. Wouldn’t want them near coral or anything.
To be clear, this isn’t an argument for underwater DCs. I just think it’s important to keep the scales of these things in view, so one doesn’t dilute actual causes of large scale climate change (GHGs etc.) with things like this that have basically zero chance of any even mesoscale effect.
I am starting to be convinced that anything we do (at scale, over time) will become "counter-environmental" ...
... it is as if the only thing that suits our poor planet is some sort of very delicate homeostasis, which we disrupt, in what amounts to, our fight against entropy.-
Couldn't you get the cooling of the ocean from just pumping water through pipes in the ocean and then onto land? And if nitrogen was a big benefit, couldn't you do that on land too?
Not knocking the experiment, it seems very interesting / worth seeing what happens.
Yeah Google built a data center in an old paper mill in Norway. They use ocean water to cool the heat exchangers. It’s important not to let the salt water in. Things get a lot more difficult then.
I wonder if they can use the same cooling tower for their nuclear plant and their data center? Colocating fission reactors and datacenters might be a way to get some use out of all those empty malls resulting from amazonification.
Hacker News is getting really useless. I thought I'd heard this some time ago, and yep, clicking through,this is indeed "news" that is OVER A YEAR OLD.
Are we just completely ignoring the "News" part of Hacker News now?
What I guess is that the thing was ok in terms of not being a giant disaster but it did not have any added value compared to a land based équivalent and the external factors costs like maintenance on the server, operating and monitoring the container, ... were terrible compared to a normal datacenter.
Good that this didn't become a trend. Oceans warming is one of the huge threats of climate change, we really don't need to help it.
Add to that it was done in the North where ocean's cooling is needed for Golf stream to keep circulating.
Don’t ever pee in the ocean … to avoid accelerating global warming
A data center is a near-zero contribution to ocean warming.
Where do you think most of the excess energy from anthropogenic climate change is being absorbed? It's measured on a scale of hundreds of 1e21 J, of which each 1 is 1 million 1 Mt nuclear bombs equivalent. We're already at +200 from 1993. Human activity is leading to massive changes in Earth's delicate systems, and data centers folks want to more than double energy production primarily for AI.
You know if you add enough small things together, they become a big thing, right?
You're not wrong, but not all values of big are equivalent.
e.g. volume of the worlds oceans is ~1e21 L and annual global energy production is ~3e16 Wh = ~1e20 J
1e20/(1e21*4e3) = 0.000025 ΔC
So even all world's energy production dumped into the ocean as waste heat is a minuscule direct effect. It's all in the second order effects of generating that energy...
What about the local effects of raising water temperature? (In the scenario where a large number of underwater data centers were built near each other)
It's generally bad and kills local eco system with algae blooms and messing up seasonal cycles of nutrients/food. But that's just with dumping the heat into environment especially in freshwater river based reactors and a pretty raw system of dumping hot water into the river ...
Clever use of physics and cooling towers allows us to mitigate the problem..release the energy slowly over time or use the thermal for something else.but costs a lot more than just dumping it into ocean and doing lazy misleading napkin math.
Yes of course, local impacts would be more pronounced. Wouldn’t want them near coral or anything.
To be clear, this isn’t an argument for underwater DCs. I just think it’s important to keep the scales of these things in view, so one doesn’t dilute actual causes of large scale climate change (GHGs etc.) with things like this that have basically zero chance of any even mesoscale effect.
I am starting to be convinced that anything we do (at scale, over time) will become "counter-environmental" ...
... it is as if the only thing that suits our poor planet is some sort of very delicate homeostasis, which we disrupt, in what amounts to, our fight against entropy.-
Couldn't you get the cooling of the ocean from just pumping water through pipes in the ocean and then onto land? And if nitrogen was a big benefit, couldn't you do that on land too?
Not knocking the experiment, it seems very interesting / worth seeing what happens.
Yeah Google built a data center in an old paper mill in Norway. They use ocean water to cool the heat exchangers. It’s important not to let the salt water in. Things get a lot more difficult then.
Not sure if they built one in Norway too, but you might mean Finland. https://medium.com/@henrihapponen/the-coolest-data-center-in...
I suspect it is similar underwater ;)
I am sure this has got to have been tried before ...
(Then again, pumping the water would consume some energy ...)
I wonder if they can use the same cooling tower for their nuclear plant and their data center? Colocating fission reactors and datacenters might be a way to get some use out of all those empty malls resulting from amazonification.
This should become industry standard. Well thought out.-
Hacker News is getting really useless. I thought I'd heard this some time ago, and yep, clicking through,this is indeed "news" that is OVER A YEAR OLD.
Are we just completely ignoring the "News" part of Hacker News now?