The Data Submarine

  • Comments posted to this topic are about the item The Data Submarine

  • Interesting, there are tales of gigantic eels living in the cooling water outflows of some old coal fired powered stations - the divers charged with the responsibility of keeping the channels clear were not happy men, and they went through quite a lot of them... Seriously though, as a physicist, I am dismayed by all this talk of 'waste heat'. Any temperature differential can be used to do work, heat a greenhouse for example (and maybe use some of the waste 'greenhouse' gases in a greenhouse, to grow food, and other cash crops) The Dutch make good money from greenhouses, and export to the world. So it can be done, and profitably. In fact, any datacentre could also be used to heat and power and feed and accommodate the homeless... Silicon Valley being a prime example, where a lot of work needs to be done to ameliorate the harm done by the greedy 1% Just a bit of systems thinking needed. All MS is doing here is dump heat into another environment that may not be able to sustain it, just a cheap and ersatz green way of doing it...

  • I assume that this is in fact a case of saving energy in terms of energy used in cooling. So it's not a case of heat being wasted, but energy being saved on cooling which has to be direct. Energy that could otherwise have been used to heat a greenhouse perhaps. I don't know this, but I'm guessing that the energy saved in cooling must be substantial enough to outweigh the risks of burial at sea!

  • walkerjian - Tuesday, June 12, 2018 10:29 PM

    ... All MS is doing here is dump heat into another environment that may not be able to sustain it, just a cheap and ersatz green way of doing it...

    No additional energy is required to remove this heat. By comparison, extensive additional is required for air conditioning cooling on land, in addition to the amount of heat disapated.

    But I think this is just a trial balloon to see how it works, not necessarily a real 'direction' at this point.

    ...

    -- FORTRAN manual for Xerox Computers --

  • I was curious about how this would actually work, were they using the equivalent of a desktop water-cooling rig (heatsink on CPUs with water flowing through to a radiator) or something else.  Reading the article, it's something else.

    Essentially, they've got radiators on the backsides of the server racks, with fans to pull the air over the servers and through the radiators.  Not as efficient as having the heatsink on the processor, but does have the advantage of sucking up the waste heat from other components on the servers (FETs, NICs, etc.)

    The big question is, is the total cost of something like this, including the installation, cheaper (both up front and over time) than a typical land-bound data center of equivalent size?  And, as others have pointed out, the warm water *will* attract sea life to the outflow, potentially blocking it, and the inflow has a chance of clogging up from pulling in debris and sea life.  Presumably they've got a screen to prevent such material from actually getting pulled into the system itself (or at least objects over a certain size,) but even so, eventually that screen is going to get covered...

  • My concern is the environment, what affect this will have on sea life?  Let's quit using the oceans as a dumping ground.  I know there is no visible waist, but the heat entering the water is 'waste'.  The oceans are warming up enough now, let's not add more to it.  As walkerjan said there are better ways of doing this.

    -------------------------------------------------------------
    we travel not to escape life but for life not to escape us
    Don't fear failure, fear regret.

  • below86 - Wednesday, June 13, 2018 8:05 AM

    My concern is the environment, what affect this will have on sea life?  Let's quit using the oceans as a dumping ground.  I know there is no visible waist, but the heat entering the water is 'waste'.  The oceans are warming up enough now, let's not add more to it.  As walkerjan said there are better ways of doing this.

    Compared to the heat from other sources, sun, rivers, geothermal vents, even ships, this is incredibly small potatoes in a very large heat sink. Effects would extend only a few feet.

    ...

    -- FORTRAN manual for Xerox Computers --

  • allinadazework - Wednesday, June 13, 2018 4:30 AM

    I assume that this is in fact a case of saving energy in terms of energy used in cooling. So it's not a case of heat being wasted, but energy being saved on cooling which has to be direct. Energy that could otherwise have been used to heat a greenhouse perhaps. I don't know this, but I'm guessing that the energy saved in cooling must be substantial enough to outweigh the risks of burial at sea!

    Cooling power is substantial. When we used to rent space for SSC, cooling was charged for first. We got power limitations first, then cooling charge, then charge by power.

    It would be better to have this heat used elsewhere, though I don't know how feasible this is. It's not likely to boil water, but perhaps greenhouse or something else. However, that doesn't necessarily help the data center survive, since it needs some cooling as well as heat dump.

  • Steve Jones - SSC Editor - Wednesday, June 13, 2018 8:27 AM

    allinadazework - Wednesday, June 13, 2018 4:30 AM

    I assume that this is in fact a case of saving energy in terms of energy used in cooling. So it's not a case of heat being wasted, but energy being saved on cooling which has to be direct. Energy that could otherwise have been used to heat a greenhouse perhaps. I don't know this, but I'm guessing that the energy saved in cooling must be substantial enough to outweigh the risks of burial at sea!

    Cooling power is substantial. When we used to rent space for SSC, cooling was charged for first. We got power limitations first, then cooling charge, then charge by power.

    It would be better to have this heat used elsewhere, though I don't know how feasible this is. It's not likely to boil water, but perhaps greenhouse or something else. However, that doesn't necessarily help the data center survive, since it needs some cooling as well as heat dump.

    In theory, I would think you could put the chillers for a datacenter inside a greenhouse of some sort, but the problem would be, this would adversely impact the ability of those chillers to cool down...
    The heat has to have someplace to go, and if they're already in a warm / hot environment, they'd have to work that much harder to cool.  Arguably, in that case, all you'd be doing is moving the energy waste from the datacenter (waste heat being pumped into the air) to the power plant / transmission lines (generating more power to run the cooling system that's working harder.)

  • jasona.work - Wednesday, June 13, 2018 7:30 AM

    I was curious about how this would actually work, were they using the equivalent of a desktop water-cooling rig (heatsink on CPUs with water flowing through to a radiator) or something else.  Reading the article, it's something else.

    I suspect there are radiators of some sort to absorb heat inside and radiate to the container. I wonder if they would have pumps at all since those typically need maintenance. Might be passive cooling piping that lets water flow, perhaps using natural movement as water heats. There isn't air inside, so no moisture issues. I would suspect that they hope it remains uncovered and water flows across it, just using convection for heat dispersal.

  • below86 - Wednesday, June 13, 2018 8:05 AM

    My concern is the environment, what affect this will have on sea life?  Let's quit using the oceans as a dumping ground.  I know there is no visible waist, but the heat entering the water is 'waste'.  The oceans are warming up enough now, let's not add more to it.  As walkerjan said there are better ways of doing this.

    The article addresses some of this, though there is a difference between 1 of these and 100 of them in any space.

    "The biggest environmental concern with a system like this is the heat it produces as the byproduct of computing. Microsoft researchers claim that the heat it puts out doesn’t make enough of a difference to be concerned. Results from the first phase of Project Natick showed that the heat from the pod quickly mixed with cold water and dissipated due to currents.

    “The water just meters downstream of a Natick vessel would get a few thousandths of a degree warmer at most,†the researchers wrote in an article for IEEE Spectrum. “So the environmental impact would be very modest.â€

    "

  • Steve Jones - SSC Editor - Wednesday, June 13, 2018 8:50 AM

    I suspect there are radiators of some sort to absorb heat inside and radiate to the container. I wonder if they would have pumps at all since those typically need maintenance. Might be passive cooling piping that lets water flow, perhaps using natural movement as water heats. There isn't air inside, so no moisture issues. I would suspect that they hope it remains uncovered and water flows across it, just using convection for heat dispersal.

    And I'm wrong. Reading closer, there is piping for move water through the interior and they are concerned about fouling. Still a chore/job that likely will be needed.

    though, perhaps AI robots can do this.

  • Steve Jones - SSC Editor - Wednesday, June 13, 2018 8:53 AM

    below86 - Wednesday, June 13, 2018 8:05 AM

    My concern is the environment, what affect this will have on sea life?  Let's quit using the oceans as a dumping ground.  I know there is no visible waist, but the heat entering the water is 'waste'.  The oceans are warming up enough now, let's not add more to it.  As walkerjan said there are better ways of doing this.

    The article addresses some of this, though there is a difference between 1 of these and 100 of them in any space.

    "The biggest environmental concern with a system like this is the heat it produces as the byproduct of computing. Microsoft researchers claim that the heat it puts out doesn’t make enough of a difference to be concerned. Results from the first phase of Project Natick showed that the heat from the pod quickly mixed with cold water and dissipated due to currents.

    “The water just meters downstream of a Natick vessel would get a few thousandths of a degree warmer at most,†the researchers wrote in an article for IEEE Spectrum. “So the environmental impact would be very modest.â€

    "

    The heat to the water may be minimal downstream for one, but as you say 100 or 1,000 of these, and heat a tracks life.  What happens to that life when one or more of these goes down?  What happens when/if they decide to remove them all together?  You have created a small eco system that will be damaged.  There is always a ripple affect.

    -------------------------------------------------------------
    we travel not to escape life but for life not to escape us
    Don't fear failure, fear regret.

  • Steve Jones - SSC Editor - Wednesday, June 13, 2018 8:53 AM

    Steve Jones - SSC Editor - Wednesday, June 13, 2018 8:50 AM

    I suspect there are radiators of some sort to absorb heat inside and radiate to the container. I wonder if they would have pumps at all since those typically need maintenance. Might be passive cooling piping that lets water flow, perhaps using natural movement as water heats. There isn't air inside, so no moisture issues. I would suspect that they hope it remains uncovered and water flows across it, just using convection for heat dispersal.

    And I'm wrong. Reading closer, there is piping for move water through the interior and they are concerned about fouling. Still a chore/job that likely will be needed.

    though, perhaps AI robots can do this.

    Heh, I was just getting ready to point out that in one of the pics in the article (and the MS blog post it links to as well,) you can see the "guts" of the datacenter, including regular air-fans to circulate the air in the pressure vessel.

    More likely, for clearing the intake / outflows, they'd use an ROV rather than an autonomous robot, if only to reduce the chance of an accident and have a better chance to deal with something unexpected (like a not-so-friendly octopus making it's home over the outflow.)  If I recall right, as well, the unit is only ~100-150 feet down which is well within sport diving (to say nothing of professional diving) limits.  In which case they could hire a local diver to go down once a month with a crowbar to clear off the intake and outflow, probably for less than an ROV would run.

  • 100 ft is fine. I've been down that far diving, and it would be a job. I hope they try humans, ROVs, and Roombas. I could see some robot that lives on this thing, circling it and keeping things clear.

    Be interesting to see how it turns out in a year.

Viewing 15 posts - 1 through 15 (of 25 total)

You must be logged in to reply to this topic. Login to reply