[Beowulf] More about those underwater data centers

C Bergström cbergstrom at pathscale.com
Mon Nov 5 12:50:13 PST 2018


Building cooling maybe.. Then again in the UK I doubt the need would be so
strong. The building from aerial view is ying/yang so it's probably just
design

On Tue, Nov 6, 2018 at 4:46 AM Prentice Bisbal via Beowulf <
beowulf at beowulf.org> wrote:

> Yes. Something exactly like that! Is that what that pond is used for? I
> would expect that is much larger than what is needed for a typical data
> center.
>
> Prentice
>
> On 11/05/2018 01:35 PM, John Hearns via Beowulf wrote:
>
> > Honestly, though, I think most of this is moot. With direct-contact
> liquid cooling and warm-water cooling, I think for most data centers,
> cooling to ambient air should be adequate. For  >places where that isn't
> enough, I would think a shallow, man-made cooling pond on premises would be
> an adequate heat sink, without having to go all the way to the ocean. By
> keeping >it shallow, at night when it cools off, the pond could dump a lot
> of its heat to the atmosphere.
>
> Something like this perhaps?
> https://youtu.be/0gCXfWCLZAA
>
>
>
>
>
> On Mon, 5 Nov 2018 at 16:01, Prentice Bisbal via Beowulf <
> beowulf at beowulf.org> wrote:
>
>> Prentice
>>
>> On 11/05/2018 06:02 AM, Stu Midgley wrote:
>>
>> As far as I can tell, they are just using the salt water to reject the
>> heat to.  How they get the heat from the cpu/hot bits to the water is not
>> clearly stated...
>>
>> A passive heat exchanger would make energy sense... but would cost a bomb
>> in engineering...  maybe direct fluid cooling (asetek) with a
>> heat-exchanger to the salt water?
>>
>> Either way, its stupid.  They could just easily pump the cool salt water
>> from the ocean into a DC, reject heat to it using the same methods... and
>> pump it back to the ocean.  Since no real delta in height, it would be
>> efficient in energy.
>>
>> The issue with this would be the increased maintenance cost of the
>> equipment pumping the salt water to the the DC, do to the corrosion from
>> the salt water, and overall 'dirtiness' of the saltwater. A better approach
>> would be to have a closed loop of treated freshwater going from the data
>> center to the a heat exchanger submerged in the sea. This should reduce
>> maintenance costs for the system.
>>
>> Honestly, though, I think most of this is moot. With direct-contact
>> liquid cooling and warm-water cooling, I think for most data centers,
>> cooling to ambient air should be adequate. For places where that isn't
>> enough, I would think a shallow, man-made cooling pond on premises would be
>> an adequate heat sink, without having to go all the way to the ocean. By
>> keeping it shallow, at night when it cools off, the pond could dump a lot
>> of its heat to the atmosphere.
>>
>>
>> OR... just use a boat...
>>
>>
>>
>> On Mon, Nov 5, 2018 at 2:27 PM <jaquilina at eagleeyet.net> wrote:
>>
>>> Probably a stupid question here,
>>>
>>> What is the advantage of using salty sea water lets say over for example
>>> mineral oil? I have seen on you tube these guys showing that a pc will
>>> still run in a fish tank and all components submerged in mineral oil?
>>> Yes it will be messier to change components but would the use of mineral
>>> oil be more efficient?
>>>
>>>
>>> On 2018-11-04 14:10, Gerald Henriksen wrote:
>>> > On Sat, 3 Nov 2018 18:27:05 +0000, you wrote:
>>> >
>>> >> I’m not sure there’s a huge population of Xcloud-Xbox gamers in
>>> >> Orkney.  There's not much daylight this time of year, of course, so
>>> >> maybe that's what those Orcadians are up to.
>>> >
>>> > Likely just a convenient place for a second test unit.
>>> >
>>> > In a way this is just an extension of the idea/product Sun came up wth
>>> > where they put a datacentre in a shipping container with the idea that
>>> > you could quickly get the datacentre where it was needed.
>>> >
>>> > While I wouldn't say this won't fail, I think there is a lot of
>>> > attraction to the concept given not just the time lag do build a
>>> > traditional data centre (mentioned in the article), but even the cost
>>> > of real estate in many/most places people live these days.  Do you,
>>> > for one example, want to pay NYC rents or just throw a bunch of pods
>>> > in the Hudson?
>>> >
>>> > I guess once you accept the idea that we no longer maintain these
>>> > datacentres in the traditional way - we now just let hardware fail in
>>> > place and ignore it until it's time to replace all the hardware -
>>> > moving to smaller sealed units doesn't seem to strange.
>>> > _______________________________________________
>>> > Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin
>>> > Computing
>>> > To change your subscription (digest mode or unsubscribe) visit
>>> > http://www.beowulf.org/mailman/listinfo/beowulf
>>> _______________________________________________
>>> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
>>> To change your subscription (digest mode or unsubscribe) visit
>>> http://www.beowulf.org/mailman/listinfo/beowulf
>>>
>>
>>
>> --
>> Dr Stuart Midgley
>> sdm900 at gmail.com
>>
>>
>> _______________________________________________
>> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
>> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
>>
>>
>> _______________________________________________
>> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
>> To change your subscription (digest mode or unsubscribe) visit
>> http://www.beowulf.org/mailman/listinfo/beowulf
>>
>
>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
>
>
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit
> http://www.beowulf.org/mailman/listinfo/beowulf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20181106/30f2ad39/attachment-0001.html>


More information about the Beowulf mailing list