Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
Hardware Hacking Hardware

Oil Soaked Servers Coming Soon 321

grease_boy writes "A UK company will start selling server racks submerged in oil baths within a year. Very-PC is working on prototypes and says that because oil transfers heat more efficiently, power usage can be cut by fifty percent."
This discussion has been archived. No new comments can be posted.

Oil Soaked Servers Coming Soon

Comments Filter:
  • Heh (Score:4, Funny)

    by peterprior ( 319967 ) on Wednesday April 11, 2007 @09:06AM (#18687819)
    Bring a whole new meaning to Oil in a rack...... geddit..

    *grabs coat*
  • Interesting (Score:5, Funny)

    by LordPhantom ( 763327 ) on Wednesday April 11, 2007 @09:07AM (#18687835)
    That sounds like a step forward. At least, until you consider that anyone working on them would get coated in oil... and frankly, server admins coated in oil are really something nobody wants to see.
  • Hurrah! (Score:3, Insightful)

    by Rob T Firefly ( 844560 ) on Wednesday April 11, 2007 @09:10AM (#18687863) Homepage Journal
    This is fantastic! I can't see a single downside to increasing the demand for machine oil in this modern world, nosirree..
    • First time I saw this it was done with cooking oil.
  • I can see how people would want to do this when hacking their own motherboard, but I would not like to see this become commonplace. For a starter, what to do with the oil after it has been used. I presume that you cannot reuse the oil to bake fries in. And I would really like to know if this would have negative implications considering the life-time of the equipment as well.
    • by leuk_he ( 194174 )
      THat oil can be reused real easy. e.g. to convert it to diesel oil to run your car on.

      The only problem i can see is that once you bath your pc components in oil you cannot reuse them elsewhere because the contacts get all dirty. Also i wonder if the components on a Motherboard can handle being oaked in oil. I can imaginge some component will solute in oil after a month or so.

      Note also that Harddrives can not be soaked in oil (they need the air cushing )
      • There are many wonderful highly toxic chemicals that can easily clean the oil off...

        But back to the main concern... Obviously when needed you would drain the oil and filter it so it can be reused for cooling again. I would assume that the system would be closed and the same oil could be used for dozens / hundreds of years.

        I would also assume that servers could be more dense - much more dense than traditional blade servers as long as you pump the oil around a little. The spacing between boards could be drast
  • So will Al Gore come out with a corn-oil version?
    • Re: (Score:2, Informative)

      by xfmr_expert ( 853170 )
      Actually, vegatable oils (natural ester fluids) have been used as an alternative dielectric fluid for several years now. A fair number of distribution-size transformers are filled with it, as it has less environmental consequence in the event of spills. It does have lower oxidation stability than mineral oil, so the system would have to be sealed.
  • Cut power in half? (Score:3, Interesting)

    by theantipop ( 803016 ) on Wednesday April 11, 2007 @09:15AM (#18687921)

    "It is possible to cut power consumption in half," managing director Peter Hopton told New Scientist. "You don't need to drive inefficient fans, or the usual air conditioning."
    Do data centers really use as much power cooling the server farms as running them?
    • by Mprx ( 82435 )
      He doesn't specify what power consumption he is talking about, so he probably means the power consumption of the old cooling system. Much less impressive than halving total power consumption.
      • Well using an oil bath to coll a semiconductor doesn't magically halve the current needed to switch a transistor so he's got something mixed up here. Or it's a bald-faced lie.
    • Re: (Score:3, Informative)

      by Sobrique ( 543255 )
      It's probably not far off. Bear in mind that a lot of the 300W of your power supplies in each system is dissipated as heat. I've got a datacenter that's had water cooled racks installed (which as you might imagine, has horrific 'overheads' on installation, cableing and maintenance). At £5k/rack, + overheads, it was still a cheaper solution than standard 19" rack + aircon bill.
      • Well, the fuzzy thing about this is that the heat still has to go somewhere. Granted you may not be powering thousands of tiny, whiny fans to remove the heat from the device, but now you've got a heated mass of oil that itself needs to be cooled off.
        • Simple. Use a radiator, and put it outside. Most of the heat can be dumped that way, and then you just need a small amount of cooling to get it down when it is really hot outside.
        • They could take a lesson from the much-maligned President Bush, and use deep groundwater (I believe he has pipes going down 300 ft at his Crawford, Texas home) to provide a heat sink. Alternately, copy Toronto, where a large pipe is being constructed in Lake Ontario that provides water at approximately 4 degree C to cool office buildings in the downtown core.

          My question is "Why do they have to bathe the components in the liquid?". I seem to recall old stereo components with finned heat sinks sticking ou

          • by Aladrin ( 926209 )
            "Couldn't some variant of this technology be used, where the actual PCB's don't get near the liquid, but the heat sinks are immersed in it?"

            I've got this great idea! Let's put the liquid in TUBES and have pumps to circulate the liquid.

            The answer is simple: That's already being done. This is only news because it sounds crazy and someone did it. It sounds crazy on the surface because 'everyone knows' that liquids and electronics don't mix. You've got to have a bit of knowledge about physics to overcome t
    • by Spazmania ( 174582 ) on Wednesday April 11, 2007 @09:34AM (#18688131) Homepage
      Do data centers really use as much power cooling the server farms as running them?

      More or less, yes. Efficiency on the A/C units is usually around 2:1 and sometimes approaches 3:1, that is you get twice the cooling as the energy you put in. Since nearly 100% of the power in to servers is expressed as heat, you need the same amount of cooling. Now add inefficiencies in the cooling architecture, power for fans in the servers, inefficiency of semiconductors when running hot, etc. When you add it all up you're approaching 50% of the total power consumption.

      Its a disingenuous marketing claim though. Cooling oil is no more efficient than cooling air and convection won't be the final word at an industrial scale - they'll need pumps which consume as much energy as fans

      On the plus side 10kva in a oil-cooled rack will be a hell of a lot quieter than 10kva in an air-cooled rack with a hundred 3cm fans running at 7krpm.
      • by DjMd ( 541962 )
        I wouldn't go so far as to call it disingenuous.

        You got it all right regarding A/C inefficiency.
        So right there the efficiency improvement means that C/kva ratio is better. (More cooling for less power)

        But the goal of using oils is their heat transfer properties. Allowing for low speed pumps and radiators. In an ideal system, it would have integration into the building. Servers in their oil, heat exchanged with some central system that leads to roof radiators.

        In Warm climates this would be less effective...
      • Since nearly 100% of the power in to servers is expressed as heat, you need the same amount of cooling.

        That's why I drive a bank of high-intensity LEDs from each server power supply, and shunt the light out of the building with mirrors. It considerably lowers the fraction of power we spend on cooling!

      • Re: (Score:3, Funny)

        by Brickwall ( 985910 )
        Efficiency on the A/C units is usually around 2:1 and sometimes approaches 3:1, that is you get twice the cooling as the energy you put in.

        In this /., we obey the laws of thermodynamics!

      • SEER = BTU / W.h (Score:3, Informative)

        by redelm ( 54142 )
        While I generally agree with the parent, AC is more efficient than 2:1 or 3:1 . An older SEER 10 AC unit is then 2.94 W [heat] removed per 1 W [electric]. The newer SEER 13 units are 3.8 W/W .

  • A sensible idea. (Score:2, Interesting)

    by 91degrees ( 207121 )
    HWSpirit did a proof of concept here [hwspirit.com]. I wonder if these guys were inspired by that.

    But it's a decent idea. Oil has a high thermal capacity and will circulate through convection keeping the temperature down. Repairs and upgrades aren't going to be all that pleasant but some swarfega will get the grease of your hands after changing the motherboard.
  • So, now you're going to have to take your computer to the garage to change the oil?
    • Re:Changing the oil (Score:4, Interesting)

      by aadvancedGIR ( 959466 ) on Wednesday April 11, 2007 @09:28AM (#18688069)
      I know it was a joke, but I fear that, by replacing air by oil, the weight of a server rack might be a problem if it is not located in the basement.

      Anyhow, even by reducing the power requirements by using efficient passive cooling to evacuate heat from the chips to the room, you still need to evacuate heat from the room.
    • by kjart ( 941720 )

      I'm not so much worried about changing the oil as much as changing failed components/servers. I'd think that could be a bit annoying, what with the tank of oil and all.

  • by account_deleted ( 4530225 ) on Wednesday April 11, 2007 @09:17AM (#18687957)
    Comment removed based on user account deletion
    • Re: (Score:3, Funny)

      by inviolet ( 797804 )

      Do you want me to make the joke about "fried chips" or do you want to do it?

      Give it to us raw -- and wriggling! You keep nasty chips!

    • Re: (Score:2, Funny)

      With the right chip you could make french fires at the same time
  • I've overheard her discussing the whole home server obsession with her friends, she is fairly tolerant of my hobby and believes it has one major saving grace.

    Its clean!

    Especially compared to say motorbikes as a hobby, oily bike parts in the sink are not a good way to endear yourself aparantly...

    But now servers come in oil? I can see problems starting here!! ;)
  • I'm sure the Slashdot editors will check back with the company (Very-PC) next year to see if they brought any oil-cooled servers to market. And report back with the results.

    Liquid cooling has been around for a long time and has advantages over air cooling. Energy efficiency is not one of them.
  • by Flying pig ( 925874 ) on Wednesday April 11, 2007 @09:26AM (#18688039)
    They will have to run the HDDs outside the oil because they do, in fact, need ventilation. Though perhaps you can get totally sealed HDDs from somewhere by now.

    However, the main problem I see is connectors. Existing connectors have been developed to work in air, except for a few exotic types. Watertight connectors are designed to work with wet environment outside and dry electronics inside, not vice versa, but in any case existing technology would require standard connectors to be used entirely submerged in dielectric. Modern connectors have much smaller contact surfaces than they did even ten years ago, and the distance liquid would have to move by capillary action before breaking the contact is quite small. It's hard to see how you could do accelerated life testing for such a system, which means it could be many years before we know whether they are reliable or not.

    I recall when doing research involving electronics in Fluorinert we had to make soldered connections in liquid. Contacts that were frequently made and broken could be pressure contacts, but that is quite different from the situation in a server. And if we had known of a cheap substitute for Fluorinert we would have used it. The majority of oils degrade quite interestingly - you wouldn't expect bacteria to live in them but they can and do if the conditions are right.

    These guys may have a workable solution to all the problems, but I can't help thinking that technology will make the concept obsolete. How does the performance of an old Fluorinert-cooled Cray stack up against a modern server in flops and GBit/s of IO per watt? (Hint: Don't bet on the Cray.)

    • One other problem that folks have had in the past with running computer components in oil is that the stuff tends to creep through cabling and leak out the distal end of the cable. Messy. That means that networking, keyboards, monitors, hard drives, etc will very likely have to be coupled to the motherboards by some wireless technology. Not a problem in a home, but is there going to be an interference and bandwidth issue with hundreds or thousands of wireless devices in a server farm?

      Probably leakproof

      • Re: (Score:3, Insightful)

        by oyenstikker ( 536040 )
        Simple. Have the outgoing cables come out of the top and connect to a patch bay, so the little oil that capillary action's itself through the cable will gravity itself right back down the outside.
    • Re: (Score:3, Interesting)

      by LWATCDR ( 28044 )
      I wonder if these servers would have to be "disposable" Trying to swap out parts is going to be a major problem. Think of the time to drain the server and get it clean enough to swap out a part. Forget about adding RAM. Probably not an issue for places that use many 1Us for a web front end but for a lot of places it seems like a big pain. I hate to bring it up but what about the fire hazard. Most oils I have see will burn if you get it hot enough.
      For everyone that was posting about hard drives I doubt tha
    • by YGingras ( 605709 ) <ygingras@ygingras.net> on Wednesday April 11, 2007 @10:59AM (#18689261) Homepage
      The design with the Cray 2 was a bit excessive. They just had the heat reach fluorinert's boiling point and there was a vapor collector and a condensor tower [flickr.com]. As you'll recall, the temperature of a liquid will not never exceed the boiling point until it all turned to vapor. That's why car are water cooled. If you have insufficient heat transfer from the radiator, the vapor pressure blows the cap and you have a really visual feedback that it's time to stop. You won't damage anything if you stop before you evaporate all your coolant. Fluorinert boiled at 56 C, a convenient temperature that makes it safe to work around the computer. Oil boils at 175 C. If you have a few boiling racks you will not want humans in your server room and you'll probably burn down your air cooled servers. Oil cooled system will not used the clever technique used by Cray: no pump or other circulatory system was needed and working temperature was ultra stable. Fluorinert and oil cooling are completely different things and I don't think you can compare them.
    • Re: (Score:3, Insightful)

      by flaming-opus ( 8186 )
      To answer your question - the last Submersion cooled cray was the mid-90's era T90, which submersed 32 processor boards and the memory, in a pool of flourinert. It did not, howerver, submerse the rest of the system, in particular the I/O, which would use standard pin connectors. One of the more exotic pieces of technology in the T90 was the robotic claw connectors that clamped down on the edges of compute boards at 400 contacts per inch. The unit costs of these connectors were very high, in part because the
  • The only problem (Score:2, Interesting)

    by xfmr_expert ( 853170 )
    The only problem is that oil is a good solvent. Of course, computer equipment is obsolete in 3-5yrs, so maybe it's not an issue. However, the article mentions they tried motor oil first, so I wonder how much they actually thought this through. Motor oil, among other things, is much more viscous than traditional dielectric fluids. The fluids used in transformers are more like water in terms of viscosity. Lower viscosity provides better heat transfer. Also, since high dielectric strength is not an issue
    • The fluids used in transformers are more like water in terms of viscosity

      I would have thought that a lighter oil like diesel or LHM (suspension oil) would have been better. Maybe even ATF, although that is a bit thicker.
      • I would have thought that a lighter oil like diesel or LHM (suspension oil) would have been better. Maybe even ATF, although that is a bit thicker.

        Or pure mineral oil. Impurities/additives generally cause oil to become more conductive, and car oils are not meant to be good insulators.

        The really good insulating oils - containing PCBs - were phased out due to their environmental toxicity in the 50s.

        -b.

  • You better put these things into the reduced-oxygen-atmosphere rooms that were mentioned a couple of weeks ago.
  • by Anonymous Coward
    • slosh

      fizz!

      Aah... not only have we saved ALL of our VAX power bill, but we've eliminated all noise in the process.

      Now to persuade the Boss to dive in to look for the drain plug...

  • Mmm...deep fried server...
  • oil (Score:4, Funny)

    by normuser ( 1079315 ) * <normuser@whyisthishere.com> on Wednesday April 11, 2007 @09:38AM (#18688187) Homepage Journal
    Oily racks you say? You know this just might catch on.
  • This sounds good until the first oil cooled computer gets slashdotted. Then the fire department has to be called. At least the computer admins will be drenched in foam forcing them to shower more out of schedule. :)
  • by DanielMarkham ( 765899 ) * on Wednesday April 11, 2007 @09:40AM (#18688219) Homepage
    It's not enough that we have jobs where we sit down all of the time, now we have a computer that's also a deep-fryer.

    Or, if they use motor oil, will Penzoil and other oil companies start running TV ads? "I couldn't play DOOM 6 until I switch to 10W-40 ultra. Now I kick butt"

    Maybe the computers can start coming with chrome valve covers.
    • It's not enough that we have jobs where we sit down all of the time, now we have a computer that's also a deep-fryer.
      If they sell one of these computers in Scotland, I guarantee you some ned will try to eat the deep-fried motherboard; it's not like there's anything we *won't* deep fry here...
  • "Hello, I'm reviewing your application and I don't see any IT exposure in it at all, no education, no experience... what makes you think that you're suited for a job at a server farm?"

    "Well, I was a fry cook at McDonalds for 2 years"

    "You're hired"
  • I always thought there were some server administrators better suited to a vocation that required them to say, "You want fries with that?"
  • by vmxeo ( 173325 ) on Wednesday April 11, 2007 @09:56AM (#18688423) Homepage Journal
    Didn't IBM use oil-cooling on one of their mini/mainframe computer systems back in the day? I seem to recall hearing stories of low oil indicators on the machines. Unfortunately my Google searches on the subject are coming up dry...
  • Why does it have to be the entire rack? Wouldn't it be easier to just use oil in a liquid radiator system and deliver the fluid to the hot spots? Sure would make it simpler to get to the computers when something fails.
  • How this will reduce power usage? The same amount of heat is going to be generated by the computer equipment, it's going to get transferred out by the oil much better than traditional air cooling, but then the oil will still be transferring the heat to the server room. It's not like the oil is making energy disappear, it's just holding on to more of it and moving it away from the computer faster. The heat will still eventually be moved to the air in the server room, which will still need to be cooled to
  • by account_deleted ( 4530225 ) on Wednesday April 11, 2007 @10:15AM (#18688623)
    Comment removed based on user account deletion
  • no, not wondering if a hot swap involves one of those deep fryer baskets.

    But with comments about doing hardware swaps and waiting for oil drainage, can't these problems be solved with keeping some components in boxes/trays, and cirulating the oil in and out of each compartment via something similar to the dry break connectors used in some motorsport hoses.

    That way you just yank it out and the only bit of oil to worry about is the amount immediately around the swapped part - which you could leave to drain, r
  • Just what we need - something else British that leaks oil. Like the MG's weren't enough. ;-)
  • All black in your face with oil, it's not just for mechanics anymore!

    Coming soon to a computer geek near you.

To invent, you need a good imagination and a pile of junk. -- Thomas Edison

Working...