Forgot your password?
typodupeerror
Power Earth IT

Companies Waste $2.8 Billion Per Year Powering Unused PCs 348

Posted by samzenpus
from the just-let-it-run dept.
snydeq writes "Unused PCs — computers that are powered on but not in use — are expected to emit approximately 20 million tons of CO2 this year, roughly equivalent to the impact of 4 million cars, according to report by 1E and the Alliance to Save Energy. All told, US organizations will waste $2.8 billion to power 108 million unused machines this year. The notion that power used turning on PCs negates any benefits of turning them off has been discussed recently as one of five PC power myths. By turning off unused machines and practicing proper PC power management, companies stand to save more than $36 per desktop PC per year."
This discussion has been archived. No new comments can be posted.

Companies Waste $2.8 Billion Per Year Powering Unused PCs

Comments Filter:
  • by jw3 (99683) on Thursday March 26, 2009 @04:59AM (#27339419) Homepage
    Of course, this solution is not for everyone, but it works quite nice at the university where I'm working. Three departments (chemistry, biology and physics) got together to form a computer administrative unit. Essentially, any workstation at one of these three dpts has the same version of OS (mostly Windows) with the same software installed. And each of these installations includes condor [wisc.edu] for distributed computing. Effectively, you get something comparable to a 1000+ nodes cluster -- and some of the machines are quite strong!
    Scientists and students alike are allowed to use it freely for their computations. There is a batch submission system, and a whole lot of numerical calculations run on these computers during night. There are a few caveats, though:
    • many biological applications need a large amount of data -- and the moment that you need to transfer gigabytes to each of the nodes (as they do not share storage) the whole thing is no longer reasonable.
    • you always have to take into account a 1-5% job loss, so if you want e.g. 1000 simulation runs, you should dispatch 1200 runs to be on the safe side. The job loss comes from a) machine being switched off b) machine having all sorts of random troubles (disk full, some weird software interaction) c) some jobs take awfully long to execute, so when 99% of your other jobs are done, you just need to kill the others.
    • Sometimes you rather launch the job locally and wait two days rather then spend half a day on preparing and testing the batch submission and get the results next morning (my time is more valuable than the CPU time...)

    All in all, you get lots of CPU, but low reliability. Which is fine for many applications. Additionally, not only you prevent energy wastage, but you also use the hardware more efficiently (so that the brand new quad core of the dpts secretary actually gets used in a reasonable way).
    By the way -- our admins hate it, when Windows computers are being switched off. They run the updates at night, as during the day the users are likely to stop an update that takes to long. I was being bashed for switching off computers during night :-)
    j.

  • Re:Same here (Score:2, Interesting)

    by Anonymous Coward on Thursday March 26, 2009 @05:11AM (#27339471)
    I'd really like to put the pc to sleep after 5 mins and to hibernation after 15.

    But give me a pc that won't die on this, and I'll do it.
  • by LoadWB (592248) on Thursday March 26, 2009 @05:11AM (#27339473) Journal

    The ROI article mentions a product which you BUY to shut down your PCs.

    I have a free solution:

    shutdown -s -t 0 -f -m

    You can schedule that at your server to force all computers to shut down at a specified time.

    Something along the lines of

    for /f "skip=3 tokens=1 delims=\" %m in ('net view') do shutdown -s -t 0 -f -m %m

    Now, you could be nice and change -t 0 to something like -t 45 and give any poor sucker at a terminal a chance to shutdown -a, or at least close programs. (There will be one error at the end for the success notice.)

    I do not recommend using that on a network without some tweaking: it will also shut down servers which show up in net view. Just a basic idea, and I do use a modified version of it at a couple of sites.

    Even a scheduled wol.exe could run to make sure computers are able to run updates overnight.

    Or you could push out a group policy that forces suspend after an hour of inactivity, and sets Windows Update to wake the computer to run. No fuss, no muss.

    Now, what did all that cost us?

  • Dumb Terminals (Score:5, Interesting)

    by Bert64 (520050) <bert@sl[ ]dot.fi ... m ['ash' in gap]> on Thursday March 26, 2009 @05:15AM (#27339487) Homepage

    Use dumb terminals, something like sunrays...

    Configure them to shut off when idle instead of run a screensaver, when you power it back on it boots pretty much instantly and the user can re-enter their password (or reinsert their smartcard) and be back where they were, all the session state is stored on the server.

    No need to keep machines on overnight for updates, because the terminals are dumb enough not to need updates...

    Dumb terminals boot instantly, so no need to keep machines pre loaded to save booting time.

    Put a power breaker by the door, last one out can turn the breaker off, first one in can turn it on (they used to do this in our computer labs at college)... There shouldn't need to be anything turned on in an office when there's no people there.

  • by fprintf (82740) on Thursday March 26, 2009 @05:53AM (#27339671) Journal

    So this was on the Reg yesterday, and the comments were all virtually the same, on two variations:

    1. The company has to pay people to sit around while PCs power up and down, eliminating any benefit from powering down the PCs since people are so much more expensive.
    2. The company pushes updates and such automatically at night when computer/network usage is low, making it less expensive (again, saving money over power saved) than pushing the updates when people turn on their computers in the morning.

    I turn most of my computers off at home and work because I hate wasting the power, and I have a problem with my home PC keeping the fan on in sleep mode. On my laptop I put it in sleep mode, plugged into the wall. I have no idea how much power this uses, but I do it so that I get a quick restart in the morning for checking slashdot @ breakfast. It bothers me that I might be wasting a few dollars per month keeping it in sleep rather than hibernate (which doesn't work on my machine - Ubuntu on a IBM T30) or full shutdown.

  • by Air-conditioned cowh (552882) on Thursday March 26, 2009 @06:02AM (#27339719)

    As a Linux user I am used to laptops and desktops never quite working because the BIOS power management only works with Windows.

    There are two possible reasons for that. One is that the open source software hasn't been written yet to take advantage of published APIs or, another possibility is that the manufacture is hiding it's APIs to make it really difficult to use anything except Windows to manage the system power.

    If it is the latter then in it seems to me highly irresponsible on the part of the hardware manufactures. How to save energy when their hardware is not being used is really not something to be hiding for any reason these days.

    I realise I don't exactly represent a significant number of users here. I'm just thinking in terms of what I can do to save energy at my own desktop (apart from the obvious switching stuff off when not in use!) and what's in the way. And Windows-centric BIOS's seem to be the main culprit.

  • by billius (1188143) on Thursday March 26, 2009 @06:08AM (#27339737)
    are malware-laden Windows boxes at small businesses with little or no regular IT Staff. I did contract IT work for small business a while back and some of the computers I had to deal with were borderline unusable. In some cases, a full reboot meant a full 15 minutes before the computer was in some semblance of working order again. That's definitely enough time to make a less savvy user want to just leave the thing on overnight and only shutdown/reboot when you really had to. And of course many of these folks didn't want to hear about how their super-awesome toolbars were the root of the problem.
  • Re:Productivity (Score:5, Interesting)

    by Zebedeu (739988) on Thursday March 26, 2009 @06:31AM (#27339825)

    Never heard of suspend? Hybernate?

  • by olden (772043) on Thursday March 26, 2009 @06:46AM (#27339891)

    Ouch. Dude, if you need to lose 15 to 20 minutes (let alone 45) to restart your PC, something is terminally wrong with your setup. Vista on a 486?
    Even in such pathological case, wouldn't suspend or hibernate be an option?

    I always power down my (work or home) PC when I expect to not need it for a while. Initiating hibernation takes me 2 seconds, resuming 30 to 40s in the rare instances when the machine is not already up again by the time I get back to it, or if I need to VPN into it.

    I'm using Linux (Ubuntu 8.10, doesn't matter much), shutting down via 's2disk'. Basically, it's hibernate, ie all applications etc are saved to disk in whatever state they happen to be, no need to exit any etc...
    s2disk uses compression by default, so while it may take a bit longer for the machine to actually finish writing everything to disk and power down (who cares), resumes are /fast/.

    Powering back up is usually triggered via the BIOS' RTC alarm, scheduled every weekday shortly before I'm expected to arrive at work. Worst case (say I'm there early), my PC is ready with all my apps running in less than 40s, time I may need anyway to check my voicemail etc.

    Remote access via my company SonicWALL SSL-VPN is also a breeze, since this gateway can issue Wake-on-LAN to whatever one wants to get to.

    Reducing waste in general is IMHO just being responsible.
    "We do not inherit the Earth from our ancestors, we borrow it from our children."

  • by LordHaart (1364019) on Thursday March 26, 2009 @07:09AM (#27339989)
    I don't think that this is in the company's best interest. $36 a year is 10c a day, and even if the machine boots in 1 minute, that's ~$20/60 = 33c of wasted employee time. So there's not that much incentive (carbon trading may change this). I'd be interested to see the effect of Sleep mode, however, as that boots much faster.
  • Re:obvious reaction (Score:5, Interesting)

    by sumdumass (711423) on Thursday March 26, 2009 @07:16AM (#27340019) Journal

    The story says $36 per desktop computer per year could be saved. Now that sound like a lot of money at a company with 500 desktops ($18,000). But that company will have at least 500 employees and probably more. At 10% more or 550 employees to work those 500 desktop computers, that brings the potential salary increases to about $32 a year. If the average person works 38 hours a week and 48 weeks a year (1842 hours), that's about a penny or less per hour raise.

    But it gets even worse. The heat cycles of computers heating up when in use and cooling down when powered off will take a small toll on the life of the computer. So I guess the real question might be is if the computer lasts 2 years instead of 3 or 4 or even 5 years, how many of those would need to be replaced because the Co2 emitted from making the things from scratch outweighed the entire carbon savings from the $36 worth of electricity not in use assuming that the power for those computers don't already come from a Co2-less generating facility. My guess is that an early replacement on any of them will offset any environmental savings which sort of makes this idea more hand waving then anything.

  • by Elrond, Duke of URL (2657) <JetpackJohn@gmail.com> on Thursday March 26, 2009 @07:27AM (#27340085) Homepage

    As annoying as all these useless background and systray apps are, isn't this as much the fault of lazy IT departments as it is the companies which produce these programs? And from all these comments, it would seem this is a problem with *many* IT departments.

    Why do the IT people leave all of this stuff on? I have to assume if people complain about it so much that they can't take it off themselves otherwise they would have long ago. So why can't IT be bothered to properly configure the machines they maintain?

    Surely most IT depts. configure one machine and ghost/clone it to others for backup and replication purposes and to prevent duplication of work. It's even less forgivable to not get rid of these apps if you only have to do it once.

  • by mokus000 (1491841) on Thursday March 26, 2009 @07:30AM (#27340101)

    I can tell you from experience in a large security-conscious organization that such pathological setups are not nearly as uncommon as you seem to think. The combination of antivirus and extremely aggressive login scripts bring fairly modern hardware with XP Pro to its knees on startup.

    When I or any of my coworkers have to cold boot, or often even just whenever we dock an already booted laptop, it means a minimum of 5-10 minutes enforced coffee break. If you're actually in a hurry to get something for someone standing in your office, it can sure seem like 30 min.

    The worst is when it boots up and tells you 10 minutes later that it's done installing some software update the login scripts had for it, so now you need to reboot. Or rather, that it's going to reboot in 30 seconds, and there's nothing you can do.

  • by mokus000 (1491841) on Thursday March 26, 2009 @07:36AM (#27340139)

    Sorry for the followup, just wanted to clarify:

    That's 5 to 10 minutes before Explorer or the start menu will respond to mouse events, not 5 to 10 minutes before the apps I need to use are open and ready to use.

    After a torturously long OS boot, I get to wait for visual studio to start up, which takes nearly as long. Add on outlook, Groove, etc, and I'd say the original poster isn't too far off on 15-20 minutes before the computer is ready to do any real work.

  • Power (Score:5, Interesting)

    by ledow (319597) on Thursday March 26, 2009 @07:38AM (#27340145) Homepage

    And any company THAT bothered by this would be using more power-efficient PC's anyway. Face it, 99% of staff using a computer as part of their daily work don't need a full desktop PC and certainly don't need dual-core systems with Gbs of RAM. So instead of faffing about trying to recoup some of the loss from buying that terrible hardware in the first place (monetary costs, environmental costs, maintenance costs, etc.) they would be much better off buying some low-power desktops (like the Atom's, Via's etc.) and thus not pumping most of their electricity into heat wastage, fans, office cooling, etc. when they could just have a small 60W or so (maximum) PC that does the same jobs.

    Those who are committed to their existing hardware - well, they should have been specifying and testing WOL, ACPI sleep, etc. in the first place if they wanted to make sure it worked in their particular environment. Chances are those stuck on old machines will have more problems trying to get the PC to sleep and to wake on cue than they would have just to buy a new cheap desktop. My pet hate is machines that won't WOL without having first been turned on manually - a power cut overnight (when the machines aren't on) means that the PC's just sit there and ignore WOL packets. And that is on fairly recent hardware (2 years old?). I know it's "wake" on LAN, but a full boot and complete shutdown (not sleep mode) will let it respond to WOL packets forever until the power disappears again.

    I would hazard a guess that the following ALL save more power than would be saved by shutting off PC's overnight for a lot less hassle and inconvenience:

    - Cutting off background services in Windows.
    - Replacing hardware with more modern equipment.
    - Disabling, centralising and/or just changing vendor of the antivirus programs to use less CPU, disk-access, etc.
    - Replacing 10% of computers with a low-power alternative (even a laptop!)
    - Turning off WAP's and other unnecessary networking hardware overnight.
    - Turning the room temperature up/down by half a degree permanently (depending on the outside environment)
    - Installing doors that shut themselves to keep hot/cold air in.
    - Opening a couple of blinds/curtains to let sunlight into some of the less-used but still heated areas (cold-countries only) or fitting blinds/curtains to reduce the heat taken in from outside (hot-countries only).
    - Training users to use shortcut keys instead of clicking the mouse for everything.
    - Or removing that poxy plasma TV in the company reception which is on permanent loop playing to nobody.

    The thing is, we take power so much for granted that when we get told to "save" it, we worry over the little bits (energy-saving bulbs) and completely forget about the larger draws (heating / cooling). $36 / year / PC is nothing, no matter the scale of the company. Even a 100 PC office (which could theoretically save $3600 / year) will probably spend multiples of that on heating/cooling, bringing someone in to do the work, or make multiples of that amount by selling off some of their old IT kit, fitting those light fittings that only switch on if someone is actually in an office, etc.

    Getting businesses to understand means providing a valid, comparable reason. That normally means *money*. But even the green-friendly companies will save much, much, much, much more money by just replacing el-cheapo PC World computer with a decent low-power one and then selling off the old kit. If you do it right, you would even MAKE money by doing this (I know it's about £200/unit for a decent mini-ITX machine, and you could easily get that for a recent second-hand machine of good spec).

    It's a *waste* of time. The proportion of power you save does not justify the effort to do it, especially not when a tiny, unnoticeable adjustment to a thermostat saves ten times the amount of power, and the hassle associated with implementing power-friendly PC's does not justify the end. Put a sign up and send a memo round to staff to turn off their PC

  • The Math... (Score:3, Interesting)

    by kenh (9056) on Thursday March 26, 2009 @07:44AM (#27340183) Homepage Journal

    The article said:

    All told, U.S. organizations will waste $2.8 billion to power 108 million unused machines this year.

    and

    By turning off unused machines and practicing proper PC power management, companies stand to save more than $36 per desktop PC per year.

    When I multiply $36 in savings per PC times the 108 Million PCs being described, I get a possible savings of $3.88B, or about $1B more than the original article reported. We "waste" $2.8B, but we can "save" $3.88B by turning off unused PCs and practicing power management? Are the savings or the waste over-estimated? One has to be wrong...

  • by Antique Geekmeister (740220) on Thursday March 26, 2009 @07:49AM (#27340197)

    One of the reasons the machines don't get turned off is the expensive 5 minutes wasted by boot times. It's an irritating waste of precious time as you return from lunch or start work in the morning, it discourages turning off boxes at night, and it discourages turning off boxes during the day when unused.

    Unfortunately, this is partly the fault of Microsoft (who enourage stupid, resource gobbling behavior at boot time like frequent resource scanning by update software and unnecessary disk indexing), and BIOS's that use ancient, proprietary, and frankly broken tools to scan for hardware that hasn't been used in 10 years. The OLPC very successfully uses a LinuxBIOS and booting procedure that cuts this lengthy pause to seconds: it should be on every server and most desktops in the country, but motherboard makers are very reluctant to support it for various reasons. As near as I can tell, it's mostly due to fear of intellectual property issues involving ancient BIOS utilities, and unwillingness to publish their own hardware knowledge associated with their own particular component selection.

    I'd love to see ASUS use LinuxBIOS by default. I've actually been asked to do that for deployments: it wasn't mature enough to use yet at that time, but it seems much more stable now and of higher quality than the average new motherboard BIOS.

  • Re:Productivity (Score:4, Interesting)

    by zehaeva (1136559) <zehaeva+slashdot.gmail@com> on Thursday March 26, 2009 @08:07AM (#27340315)
    even if you assume 1min per day to turn on you've got 20min per month. depending on your pay grade you maybe at that 36 dollars a month(200k/yearish). if you throw in having to download a profile from the server and security scripts running, even a an increase in that time of getting the computer to a usable state, say 3min per day, nets us 60min to month so that expands to people making err 70k a year? Automated off and on systems to prime the systems before the employees get in would be best. but how much to develop that? cost/benefit analysis always interests me.
  • Re:Magic smoke (Score:3, Interesting)

    by hairyfeet (841228) <bassbeast1968@gm[ ].com ['ail' in gap]> on Thursday March 26, 2009 @09:23AM (#27340881) Journal

    Well, where I am at we get our power from a pair of nuclear reactors so I ain't worried about this too much. Couple this with the fact that one of my more popular services with my SMB customers is to set it up so they can log into their office machines from home if they don't feel like coming into work but want to get a little work done, and the fact that WOL can sometimes be iffy and having a PO'ed client bitching at you because he had to come into the office after a dentist appointment because for whatever reason WOL didn't actually wake, and you can see why I just advise them to leave the thing on.

    Finally the PC I am typing this on is going on its 9th year without so much as a PSU change(only thing that failed was a HDD, and I killed it with overwork, pure and simple) and right next to it is a 5 year old gaming PC that I turn off nightly because of the fans, and it has gone through a mobo,PSU AND a HDD, and I think I'll just stick with leaving them on, thank you very much. I know that anecdotes aren't evidence but I have seen that my customers that leave theirs on simply don't need the amount of work that those that turn it off do. The ones that leave it on usually only ever need a HDD whereas I see a LOT of dead PSUs and mobos from the ones that cycle down nightly.

    And since my customers are using their PCs to make money, having it work when they need it, even if that need comes up at home and the office is 25 miles away, is worth the extra expense to them. Of course getting our power from nuclear means its cheaper so that could be part of the reason they don't care. But in the end with my customers and I it all comes down to money. And if you figure in the amount of gas they save by not coming in when they don't have to, plus the extra waste that would have ended up in a landfill from parts dying due to cycling, I bet we are getting pretty close to break even. But just like that new solder that I truly believe will cost us much more environmentally due to the waste generated by failing parts than it save due to lead, so too do I believe that turning off PCs nightly will cost more in dead machines clogging up landfills than it will save by not using juice. I'd at least want to see some REAL numbers before I scream the sky is falling.

  • by arbitraryaardvark (845916) <gtbear.gmail@com> on Thursday March 26, 2009 @12:21PM (#27343639) Homepage Journal

    Once electricity prices start seriously ramping up (which they inevitably will), companies will be giving their utility bills a lot more scrutiny.

    I suggest you look at photovoltaics and a little thing called Moore's law. Electricity will be cheaper in the future than in the now. Of course, conservation, including automagically turning off unused computers, is still the best buy. But my computer stays on. 8 months out of the year, it's an efficient electric heater. 2 months out of the year, it's an undesirable heat source and i might turn off the monitor at night if i think about it. The article, not that i read it, didn't seem to account for the heating value of the "wasted" electricity. In most of the country most of the time, electric heat is a pretty good buy compared to natural gas, coal, oil, firewood, personal nuke, whale oil, or other options.

<<<<< EVACUATION ROUTE <<<<<

Working...