3 ms·

> Google has announced plans to do the same. Citation needed? IMHO putting a datacentre under water is a horrible idea, why not pump the water 10 meters higher

> Google has announced plans to do the same. Citation needed? IMHO putting a datacentre under water is a horrible idea, why not pump the water 10 meters higher and cool a DC near the shoreline?

abhiminator 8y ago
>why not pump the water 10 meters higher and cool a DC near the shoreline? Google's already doing that -- http://www.datacenterknowledge.com/archives/2010/09/15/google-using-sea-water-to-cool-finland-project http://www.datacenterknowledge.com/archives/2010/09/15/googl... As for the Google's undersea data center plans, I'm unable to find a citation for you at the moment, but I'm fairly certain that it's in the works.
- tuukkah 8y ago
  That sea water cooling DC recently became available as the GCP region europe-north1: "Our high-tech cooling system, which uses sea water from the Gulf of Finland, reduces energy use and is the first of its kind." https://cloud.google.com/about/locations/finland/ https://cloud.google.com/about/locations/finland/
jandrese 8y ago
Exactly. Full submersion seems to add a lot of complication for little gain over just submersing your heat exchangers. I guess you can avoid some moving parts (and heat generation) by letting wave action circulate your cooling fluid, but that's a small fraction of the energy needed to run a datacenter. The tradeoff is the need for watertight seals all around your datacenter (including the doors and cable access), dealing with corrosion over a larger surface area, accessibility problems, dealing with marine life (barnacles!), anchoring it against storms, environmental impact, and so on. IMHO it is a silly idea and won't prove practical in the long run.
Const-me 8y ago
Continuously pumping sea water is expensive to maintain, the primary reason being marine biology. I once saw mussels grew inside a boat engine. Second reason, land on shorelines is expensive. Many people dream about living on a seaside (except for people who did, practically speaking the idea is not that great, humidity, noise). Also many other businesses require shorelines, beaches, marinas, ports.
- jopsen 8y ago
  Pumping water from a lake seems more reasonable..
rbanffy 8y ago
Or just put it somewhere that's very dry and very windy and use air-cooling? There are a couple of cold deserts that fit the bill that aren't too far latency-wise from users and are nice for solar power.

> Google has announced plans to do the same. Citation needed? IMHO putting a datacentre under water is a horrible idea, why not pump the water 10 meters higher and cool a DC near the shoreline?

abhiminator 8y ago

>why not pump the water 10 meters higher and cool a DC near the shoreline? Google's already doing that -- http://www.datacenterknowledge.com/archives/2010/09/15/google-using-sea-water-to-cool-finland-project http://www.datacenterknowledge.com/archives/2010/09/15/googl... As for the Google's undersea data center plans, I'm unable to find a citation for you at the moment, but I'm fairly certain that it's in the works.

tuukkah 8y ago
That sea water cooling DC recently became available as the GCP region europe-north1: "Our high-tech cooling system, which uses sea water from the Gulf of Finland, reduces energy use and is the first of its kind." https://cloud.google.com/about/locations/finland/ https://cloud.google.com/about/locations/finland/

jandrese 8y ago

Exactly. Full submersion seems to add a lot of complication for little gain over just submersing your heat exchangers. I guess you can avoid some moving parts (and heat generation) by letting wave action circulate your cooling fluid, but that's a small fraction of the energy needed to run a datacenter. The tradeoff is the need for watertight seals all around your datacenter (including the doors and cable access), dealing with corrosion over a larger surface area, accessibility problems, dealing with marine life (barnacles!), anchoring it against storms, environmental impact, and so on. IMHO it is a silly idea and won't prove practical in the long run.

Const-me 8y ago

Continuously pumping sea water is expensive to maintain, the primary reason being marine biology. I once saw mussels grew inside a boat engine. Second reason, land on shorelines is expensive. Many people dream about living on a seaside (except for people who did, practically speaking the idea is not that great, humidity, noise). Also many other businesses require shorelines, beaches, marinas, ports.

jopsen 8y ago
Pumping water from a lake seems more reasonable..

rbanffy 8y ago

Or just put it somewhere that's very dry and very windy and use air-cooling? There are a couple of cold deserts that fit the bill that aren't too far latency-wise from users and are nice for solar power.