Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Renewable Energy for the Data Center?

Posted by Cliff on Sat Feb 17, 2007 12:30 AM
from the every-erg-saved-is-precious dept.
rohar asks: "The ISP/Carrier/Colo company I work for has just announced a new 'green' program. Although this is a step forward, they don't have a comprehensive environmental sustainability plan. I have been leading an open renewable energy project and I think we have 2 novel ideas for scalable and reliable renewable electrical power, the Solar Ammonia Absorption Convection Tower and the Compressed Air Wind Electrical Generation System. Do you have new ideas (Solar PV has been done, for example) for renewable power generation and conservation for the data center and other areas of industry?"
+ -
story

Related Stories

[+] News: Data Centers Expected to Pollute More Than Airlines by 2020 322 comments
Dionysius, God of Wine and Leaf, writes with a link to a New York Times story on a source of pollution that doesn't leave contrails: "The world's data centers are projected to surpass the airline industry as a greenhouse gas polluter by 2020, according to a new study by McKinsey & Co. ... [C]omputer servers are used at only 6 percent of their capacity on average, while data center facilities as a whole are used at 56 percent of peak performance." Data centers, though, might have more options for going green than airlines do, given present technology.
This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • by SubliminalVortex (942332) * on Saturday February 17 2007, @12:34AM (#18048334)
    Just a side note... Sometimes the companies need to look internally in order to put things into perspective. For example, how often does one go to the 'printer' to only find abandoned print-outs that someone, who really cared to have that information printed needed? I've seen reams of paper wasted on not only forgotten revisions of documents, but also those who print out travel directions, local restaurant coupons, etc. If you're going to save energy, keep others from wasting it. :)
    • Also,
      Why exotic energy sources?
      Your DC already has diesel generator infrastructure right?
      Burn biodiesel or straight veggie oil! it's carbon negative (the fuel represents only a portion of the carbon in the plant, the rest of which is used for feed).
      -nB
  • by WoTG (610710) on Saturday February 17 2007, @12:44AM (#18048388) Homepage Journal
    I've wondered why most A/C units for buildings seem to rely on radiating heat via waste air. There must be a decent use for that heat nearby (at least in densely populated areas). Find a pool that needs heating, or help run the hot water tanks, or create a public hot shower or something!

    Water is a much better conductor of heat than air anyway.
    • "I've wondered why most A/C units for buildings seem to rely on radiating heat via waste air. There must be a decent use for that heat nearby (at least in densely populated areas). Find a pool that needs heating, or help run the hot water tanks, or create a public hot shower or something!"

      Great idea! But, practically speaking they are not put into place in most current cases. The piping, controls, and maintenance don't make the venture FINANCIALLY sound in our current schema. If your (and many others') i
      • The thing is for areas where you can get the land you it is practical to do something similar to this. It's called Geoexchange or a Geothermal Heat Pump [wikipedia.org]. Alternatively, if you are near a major body of water you can use it for cooling (by heat exchanger of course, which I would say is simpler than a phase-change AC system).
        • Re: (Score:3, Interesting)

          Alternatively, if you are near a major body of water you can use it for cooling

          Good luck getting your local building department to agree with you on that one. Typically using 'public' bodies of water for anything practical like this will get you run out of town. In New England, we would love to start using Geothermal systems on our houses with an ocean source or water well as the heat sink, but some towns prohibit wells of any kind, and the DEC's conservation commission has took us to task on laying pi
      • I believe that the financials on this will start changing in the near future. Hosting is so competitive and as the environmental focus increases, I think that an environmentally responsible data center will start to be the deciding factor in a customer choosing a hosting provider. Most of the other factors are equal across data centers and the customers are starting to implement their own environmental plans.
    • Why not combine your server farm with a tomato farm and use the surplus heat for the greenhouses. It only works in a cold climate but...
    • I know one Canadian data center that uses the waste heat to heat the building during the winter.

      They set this up in the mainframe days, but I'm sure today there are racks of Intel boxes next to the mainframe helping to heat the building.
    • It becomes a question of whether spending the money putting the heat to use saves more money than doing things to produce less heat(i.e., more efficient windows, better insulation, more efficient computers, replacing the air conditioner sooner, etc.).
    • I've wondered why most A/C units for buildings seem to rely on radiating heat via waste air. There must be a decent use for that heat nearby (at least in densely populated areas). Find a pool that needs heating, or help run the hot water tanks, or create a public hot shower or something!

      Because the waste heat is very low grade - it's diffuse and not greatly above local ambient. By the time you add the needed piping, pumps, heat exchangers, etc... you've added greatly to the cost - but won't be able to not

      • you've added greatly to the cost - but won't be able to noteably dimish energy needs without creating a system with a payback time on the order of several decades.

        That reply sums up both the problem and the answer. Very few companies or individuals are willing to look beyond next quarters profit statement.
  • by davidwr (791652) on Saturday February 17 2007, @01:18AM (#18048564) Homepage Journal
    How much energy is lost by having AC-DC converters for battery backup only to convert it back to AC then back to DC in the server?

    Put a single pair of load-sensitive, redundant power supplies on each rack and run DC to every device. One of these should have battery backup.

    Yes, there will be a lot more wires but it will be a lot more efficient and have lower air-conditioning costs.

    Speaking of air conditioning, if you can channel the heat to something useful, that's a plus.
    • I'm not sure the enery is converted so many times there. It would be way more cost efficient to build a device that watched the electricity going though the line and subvert it only when neccesary. Such a device would only pass AC though it without the conversion in normal modes and also charge a small bartery. When the transient voltage suppresor detects too much or not enough voltage, it shuts the mains off and allows the inverter to pass energy though from the battery. The inverter is probably constantly
      • 48V (Score:2, Insightful)

        After reading the other replies, the solution "seems" obvious. Of course, that means it probably won't work.

        Offer rack-servers with power supplies that use -48V DC input. For the more common form-factors, e.g. ATX and a few others, these should be produced in enough volume to be "commodities" and priced accordingly, at least in as much as any non-consumer component can be considered a commodity.

        The parts cost to produce 100,000 power supplies that run on -48V DC has got to be less than making the same 100
        • And unfortunatly, just like truly efficient green alternative fuels, It will take a decade or so to implement on a large scale because the life cycles of existing machines are around 5 to 10 years on average. At least from my experience anyways, with upgrades and all.

          • What I meant was that any rack-mount server, or any PC re-purposed as a server and used in a server farm, could have its PS replaced by one that used 48V input.

            For ATX-sized PSes and other common form-factors, there should be enough demand to make these as a commodity and perhaps even as a vendor-supplied option.

            For proprietary form factors you won't have the efficiencies of mass production but they should still be an option.

    • by Bios_Hakr (68586) <xptical AT gmail DOT com> on Saturday February 17 2007, @04:12AM (#18049290) Homepage
      A big problem is that a PC uses several voltages and most DC systems only produce 48vdc. I'm sure you could rig a busbar and a ATX power harness with some properly-sized resistors to drop the voltage to 12v, 5v, and 3.3v. However, it'd be a nasty hack.

      Why don't we see more -48vdc powered PCs?

      Where I work, we have commercial power and generators feeding an auto-transfer panel. The output of that feeds a rectifier which, in turn, feeds a big DC busbar with a bank of batteries attached to it. The busbar has a lead that goes to every rack in the comm center. Every rack has -48vdc powered cisco gear or an inverter to power PCs.

      The AC to DC conversion is very efficient. The DC to AC conversion is very inefficient. However, the system never loses power.
      • Why don't we see more -48vdc powered PCs?

        Check out Rackables [rackables.com] , be warned ahead of time they are rather expensive (but also pretty sexy) servers. I had the luck of working with them a few years back on a cluster, and their engineers have some really cool ideas.

        For a while, I worked for these guys [lr-inc.com] , quite often we would be called in to do campus wide lighting retrofits to offset the cost of server rooms that were getting hungrier by the day. At least while I was there, no "magic bullet" to reduce costs in the

      • Which makes me wonder why there isn't a 48VDC input ATX power supply on the market. Voltage regulators are cheap compared to the rest of the components in a power supply, and many can use in excess of 48V input.
    • by rohar (253766) * <bob.rohatensky@sasktel.net> on Saturday February 17 2007, @06:56AM (#18050028) Homepage Journal
      There is a eweek article [eweek.com] on DC power in the computer room saving 15%. On the POTS [wikipedia.org]/PSTN [wikipedia.org] side of the company everything is 48v DC. I think the AC data center is more due to vendor hardware availability and the expense on non-commodity power supplies. There is convergence of the PSTN and IP networks happening in the industry and maybe there will be more on the Server and POTS equipment. I think currently the systems are separate and Telco/Hosting companies maintain both the 48v DC supplies for the POTS/PSTN and the AC supplies for the IT systems.
  • This application is a bad match for renewables unless you're in an area with really steady winds.

    Conservation might be the better tack. DC power distribution? Watch out for voltage drops though. Virtualization? Fewer but larger hard disks? There will be tradeoffs but there should be real room for improvement at the design stage.
    • What about geothermal and hyrdoeletric? In areas where they are available they are capable of provideing a constant power source. Of course the areas where these can be found are not necessarly the best places for a datacenter though...
  • Please don't abuse the term "open", it's the 21st century version of the "e-" prefix.
    • Re: (Score:1, Interesting)

      by Anonymous Coward

      This is simple physics!

      1) Wind does not create enough potential kinetic energy to serve as a useful coal/petrol replacement.
      2) Current solar conversion systems do not have anywhere near the efficiency to serve as a useful coal/petrol replacement.
      3) Even if we get "green" on energy usage the requirements of running civilization will continue to escalate rapidly.

      We need to concentrate on reality here! Yes we need to cut down pollution to stop greenhouse warming but pretending warm fuzzy solutions will work is

      • The problem is that Greenpeace, the Sierra Club, and all of their friends (namely the coal industry, ironically) put the kibosh on nuclear power when it had a chance to take over. Now, the regulations are so absurd that it cannot grow at an affordable rate in the U.S. Whatever Greenpeace blathers about now, they did a tremendous amount of damage to the nuclear power industry. So you're partly right - even after we discard your unsubstantiated refutals of the parent's #1 and #2 (there are serious efficien
      • Yup CANDU is great. No other system is better for proliferation - that lovely easy reload without shutdown makes plutonium production so painless.
      • Bah!
        "Greenpeace has always fought - and will continue to fight - vigorously against nuclear power because it is an unacceptable risk to the environment and to humanity. The only solution is to halt the expansion of all nuclear power, and for the shutdown of existing plants."

        There may be individuals in Greenpeace that don't fear the technology, but the organization's official stance is summarized above. http://www.greenpeace.org/international/campaigns/ nuclear [greenpeace.org]
    • If you can get a reputable publication to publish such a thing, including vendor names and budget analyses, I'll be happy to read it. Otherwise, stop wasting our time.

      The idea that nuclear is the cure for all evils, including bad breath and ugly sheep is somethine we were supposed to have outgrown in the 1950s when people realized that "meterless power" based on nuclear energy was a pipe dream.
  • So if your datacenter is like a lot of datacenters, you probably have a bunch of servers. Servers tend to try and be redundant. i.e two mirrored hard drives, two power supplies, second machine to fail over to. What if, instead, you hosted on virtual machines & cut down on the number of hot-failover components, and machines. You could have a smaller number of servers functioning as failovers, and you could replace parts as needed without their using up additional power....

    just a thought.
    • Virtualization is starting to happen. We have somewhere around 1000 servers in the data center and due to different business units, service levels, OS requirements, etc, the applications are deployed on separate servers, but the utilization on many of them is very low. They are planning to move many of these systems to virtual servers.
      Our previous High Availability database environment was Active/Passive clustering (HACMP or MC/ServiceGuard) where there was an entire failover server sitting idle in case of
  • Biomass (Score:3, Interesting)

    by RAMMS+EIN (578166) on Saturday February 17 2007, @04:54AM (#18049468) Homepage Journal
    How about:

      - Grow plants (or algae) that have a high energy content and do not need good land. E.g. algae can be grown in salt water, switchgrass grows well in a variety of circumstances that other crops don't like, and kudzu grows whether you want it or not.

      - Use the energy captured in these plants to generate electricity. There are various options here, too: extract oil from algae and combust that, convert the sugars in the plant matter to ethanol and burn that, or perhaps burn the plants whole.

    I don't know which combinations of growing, harvesting, extraction, conversion, and combustion are most efficient. However, there is definitely a lot of variation here. Studies that have been done that prove some combinations aren't worth it (you have to put in more energy than you get out), and others are (you can actually supply the US with enough energy, without running out of space to grow crops for food).
    • Re: (Score:3, Informative)

      There was about 25 years of research into algae done by the U.S. DOE, as well as some renewed interest from MIT. I put a page together [energytower.org] with a drawing of the idea.
    • - Grow plants (or algae) that have a high energy content and do not need good land. E.g. algae can be grown in salt water, switchgrass grows well in a variety of circumstances that other crops don't like, and kudzu grows whether you want it or not.

      - Use the energy captured in these plants to generate electricity. There are various options here, too: extract oil from algae and combust that, convert the sugars in the plant matter to ethanol and burn that, or perhaps burn the plants whole.

      Remember, th

  • Dupe and duped. (Score:4, Interesting)

    by DerekLyons (302214) <[fairwater] [at] [gmail.com]> on Saturday February 17 2007, @10:26AM (#18051310) Homepage
    This is nothing more than a slashvertisement for Rohar's crackpot 'green' energy schemes. (One of which was recently debunked [slashdot.org] on Slashdot.)
    • Slashvertisment and crackpot or not, I don't know. But what bothered me about the initial post, which I haven't seen addressed, is the search for 'novel', saying 'PV had already been done.' So, is there something WRONG with PV or the other ways that have been done, or are we just chasing novel for novelty's sake?

      So I don't know if it's poor posting of his intentions, or poor intentions.

      • There is nothing "wrong" with the existing renewable energy sources, the ones that are in use all have merits. There are also drawbacks in reliability (i.e. direct sunlight or windy days only), location dependence, non-renewable construction materials and high energy input of construction.
        As an example: In my location in Canada, the solar isolation in the summer is relatively high at around 5kWh/day due to clear skies and long daylight hours at the 52nd parallel. But, in winter we have only 8 hours of day
      • I suspect he doesn't actually work for an ISP - but is more interested in adding to his 'featured in the media' page and getting attention for his schemes.
    • ouch
      1. If you have something specific that is incorrect in the calculations [energytower.org] or economics [energytower.org], please inform me.
      2. The energytower.org project is not-for-profit and is about collecting ideas on a scalable and location independent renewable energy source, presented to draw positive input from others. All of the individual components of the system are in use commercially, but not in a complete system.
      3. Many engineers and physics experts have looked at the energytower idea. Negative critism from a /. reader didn't RTFA
      • If you have something specific that is incorrect in the calculations or economics, please inform me.

        Any engineer worth his salt will tell you that there is a vast difference between calculations on a simplified and idealized system and actual design calculation on an actual design. Therefore, since you don't have an actual design, your calculation are meaningless for serious evaluation. Your economics section contains nothing but airy handwaving - no numbers. Therefore, it is meaningless.

        The ene

  • Summary (Score:5, Interesting)

    by rohar (253766) * <bob.rohatensky@sasktel.net> on Sunday February 18 2007, @06:06AM (#18058750) Homepage Journal
    To sum up the comments:
    Conservation Ideas
    • D.C. rather than A.C. power mains
    • Waste heat recovery for structures or cottage industries
    • Power saving features in server hardware
    • Server Virtualization
    • Better High Availability/Redundancy resource management
    Generation Ideas
    • CHP (Combined Heat and Power) plants
    • Bio-fuel backup generators
    • Wind turbine and battery systems
  • This is a classic slashvertisement - a longwinded summary cut right from TFA, then a silly open question at the end "Would you like to know more?"

    Let this guy DIAF, with Roland, and Eugenia, Dvorak, and the rest of the zeros.
  • Keep it simple: (Score:3, Interesting)

    by GWBasic (900357) <slashdot.andrewrondeau@com> on Tuesday February 20 2007, @03:48PM (#18087156) Homepage

    I suggest keeping a "green" server farm simple by outsourcing your "green".

    Where I live, (Santa Clara, in Silicon Valley,) I buy solar and wind power directly from the grid. It's not the cheapest electricity, but it is affordable. PG&E, a major electric company in CA, has very low carbon output per kilowatt hour. They also allow you to sponsor reforestation, thus allowing you to recapture the carbon generated from running your servers.

    It is also possible to buy carbon credits. This is where you essentially pay someone to remove carbon from the air. At the consumer level, Terrapass [terrapass.com] allows consumers to purchase carbon credits.

    • I really wanted to mod this, but I feel obliged to respond. Green is that last possible thing you should every try to outsource. It defeats the whole purpose of green. Being green is about taking personal responsibility to reduce the total load of industry on the environment. Paying someone else to do it for you effectively allows you to ignore the core concept for which you are striving. You're essentially saying that it's too much bother for you to do it, so you're going to pay someone else. I think of it
      • I really wanted to mod this, but I feel obliged to respond. Green is that last possible thing you should every try to outsource. It defeats the whole purpose of green. Being green is about taking personal responsibility to reduce the total load of industry on the environment. Paying someone else to do it for you effectively allows you to ignore the core concept for which you are striving. You're essentially saying that it's too much bother for you to do it, so you're going to pay someone else. I think of it