Forgot your password?
typodupeerror
Technology

Environmental Concerns for a Server Room? 78

Posted by Cliff
from the external-worries dept.
christian_thoma asks: "My company is currently in the preparation phase for building a huge new manufacturing facility. While reviewing the site plan, I've discovered that there is both a cell phone tower and high voltage lines within 100-150 meters of where the server room is going to be placed. Do I need to be concerned? Are there any special considerations when designing my server room that I need to be aware of? Has anyone else had to deal with a similar situation?"
This discussion has been archived. No new comments can be posted.

Environmental Concerns for a Server Room?

Comments Filter:
  • by MerlynEmrys67 (583469) on Wednesday May 26, 2004 @01:47PM (#9260493)
    When the electrical lines fall into your server farm setting them on fire

    Good prior planning

  • by AtariAmarok (451306) on Wednesday May 26, 2004 @01:49PM (#9260517)
    I, for one, avoid using old-fashioned coal-fired Godbout and IMSAI servers in the server room. This keeps the IT operation "clean and green" and prevents global warming.
    • Those IMSAI's felt like they were burning coal, what with the red-hot LM7805 regulators glowing on the sides of the board.

      You could tell the time-in-service of a memory board by how dark the printed circuit board material had become around the regulator.

      The good old days....
  • No and no. Absolutely not an issue.
    • That's about informative as

      Yes and yes. Absolutely is an issue.
      • Except that I'd be right and you'd be wrong.
        • Except that we have no basis but your word to take your statement on, so you're not right until you can show it in some way.
          • Except that we have no basis but your word to take your statement on, so you're not right until you can show it in some way.

            Of course you're right.

            But how would you know whether I was giving accurate information or just the half-thought-through ancdotal silliness that gets modded up as insightful or informative.

            Oh wait, that's how the game is played here on /. so I'd better just shut up and play along...
            • But how would you know whether I was giving accurate information or just the half-thought-through ancdotal silliness that gets modded up as insightful or informative.

              Epistemological roadblocks aside, it's called "trust" - the reader decides how much of it they want to lend you, based on their ability to understand your thought process, and who you are.

    • Only when you don't want to have any kids...

  • FCC Part 15 (Score:1, Funny)

    by Anonymous Coward
    You do have FCC part 15 tags on all of the equipment, right?
    • The parent post would actually be informative if he had said CE tags.

      Part of the CE cert proceedure is testing RF immunity, something that isn't required by the FCC.
  • by Anonymous Coward on Wednesday May 26, 2004 @01:53PM (#9260556)
    The CIA can spy on you using electromagnetic radiation from the powerlines to irratiate your thoughts which can be picked up from the cell tower. A tin foil hat may prevent this.
    • by Anonymous Coward
      Tin foil hats are so old school. Real men wrap their head with a lead bucket.
      • No, that gives you a single point of failure -- if the signals get past the lead bucket, you're in trouble. What you want to do is have defense in depth -- eat the lead and spread is throughout your body.
        • No, you would still be just using lead, ideally you would wear the tin foil hat and then place the lead bucket over it thus giving two layers of protection .... :)

          t

          • with the relitively low melting point of lead, you could just dip your head in molten lead.
            assuming you can still move, then put the foil hat over that... then seek immediate medical attention... then seek immediate psycological evaluation.
  • by advocate_one (662832) on Wednesday May 26, 2004 @01:57PM (#9260589)
    you'll always have five bars of signal strength... or would that be the end of the old "no signal" excuse favoured by those who're trying to avoid being dragged off to the next fire-fighting fix episode...

    Seriously... get the server room shielded with wire mesh built into the walls and conductive film on the windows... like a Faraday Cage [gla.ac.uk]... then you won't get weird interference problems

    • Lead shielded drywall is what is used for doctors and dentists offices that want to be protected from X-rays. Blocks lower RF frequencies as well.
    • get the server room shielded with wire mesh built into the walls and conductive film on the windows..

      That won't block the radiation from the cellphone tower. Try wrapping your telephone in aluminum foil, just the small hole that you need to see the signal strength indicator is enough to get 5 bars of signal strength.

      I don't think that either the cellphone tower or the power lines will cause lots of interference. As others pointed out, the cellphone is too far away to be of concern. You standing next to

  • by UserChrisCanter4 (464072) * on Wednesday May 26, 2004 @01:59PM (#9260607)
    Junk science has shown that these servers will have quadruple the risk of developing cancer.

    Make sure to have the company doctor check them out occasionally, although the servers should be aware that the company doctor does not work for them.
  • Too far away... (Score:5, Informative)

    by DjReagan (143826) on Wednesday May 26, 2004 @01:59PM (#9260615)
    Inverse square law means the electromagnetic fields given off by both of those will have dropped right down to negligible amounts over the 100-150meters. You really shouldn't worry about it.
    • How close do I need to be to start worrying?
      • How high up the cell phone tower do you propose to build the server room?

        Seriously, unless you're using WiFi links between your servers, you're putting servers in the service building for the tower, or you're building the data center within a few feet of the top of the cell tower, you won't have any problems.

        Further, the metal cases of the servers function as very effective Faraday cages, so even if you were to set a server on top of the cell tower, the only risk would be that you get radio burns from acc
      • Its based on the inverse square law: x/(r^2)

        (x being a constant based on the amount of power being emitted, and r being the radius distance you are from the source)

        Well, say you measure the amount of RF/EM radiation coming from the cell phone mast at a distance of 1 metre. At 10 metres, you'd only have 1% of that. At 50 metres distance, you've only got 0.04%. At 100m you've dropped to 1/10000th (.01%) and at 150m its down to 1/22500th (0.004%)

        As for how close you have to be before you start worrying?
    • Re:Too far away... (Score:4, Informative)

      by menscher (597856) <menscher+slashdot.uiuc@edu> on Wednesday May 26, 2004 @05:52PM (#9262845) Homepage Journal
      Inverse square law means the electromagnetic fields given off by both of those will have dropped right down to negligible amounts over the 100-150meters.

      Too bad the inverse square law doesn't apply for the high voltage power lines. It will just fall off as 1/r (where r is your distance from the lines), not 1/r^2. You're right about the cell phone tower, though.

      Oh, and yes, IAAP (I am a physicist.)

      • Interesting - what is it about the power lines that makes them behave differently?

        (trying recall something about that from his undergrad physic clases, but realising they were too long ago, and i've sufficiently pickled my brain too many times since then)
        • Re:Too far away... (Score:3, Informative)

          by MrWim (760798)
          The inverse square law applies to point sources, you can imagine that the points which are equidistant from a point is a sphere, so all the radiation is shared out over the surface area of the sphere 4r^2, so the intensity is proportional to 1/r^2. Power lines are lines, so the equidistant points form a cylinder, surface area 2rh, so the intensity is proportional to 1/r. HTH
      • Too bad the inverse square law doesn't apply for the high voltage power lines. It will just fall off as 1/r (where r is your distance from the lines)

        Almost right... It would be right for a single wire, but here, you have multiple wires, half of them carrying positive voltage and current, and the other half of them a negative voltage and current. These will partially cancel out, giving you 1/r^2 instead of 1/r (for point sources it would be 1/r^3 instead of 1/r^2).

        Note that the above is for field strengt

  • Shield your ethernet cables before sealing them behind drywall. That should be obvious to the installer, but you can never be too sure. Otherwise, it's unlikely that either of those will cause a problem, since they operate at far different frequency ranges.
    • "Shield your ethernet cables before sealing them behind drywall."

      What you seal behind drywall is supposed to be conduit, nice fat conduit. Conduit with an inner diameter about twice what you think you're going to need.

  • by shrubya (570356) on Wednesday May 26, 2004 @02:11PM (#9260735) Homepage Journal
    Your servers are safe. Heck, all that EM [wikipedia.org] will shield them from Tempest snooping.

    However, if you have any fillings in your teeth, get them replaced with non-conductive material [google.com] by a certified herbalist. Or do I mean certifiable?

    Ever hear of the Inverse-Square Law? The cell phone on your belt clip pumps more EM into your precious racks (and obviously into you) than that tower.
  • Precautions (Score:5, Informative)

    by Ropati (111673) on Wednesday May 26, 2004 @02:17PM (#9260819)
    I wouldn't be overly concerned with the RF interference being generated at the site, but you should review the matter.

    First, I would suggest that you have the site tested for RF levels. The tests should be fairly inexpensive. From the results you can decide if precautions are required.

    You still need to consider future RF issues in case the cellular company decides to upgrade. It would be prudent to have some RF grounding brought into the server room and extend the ground to the racks and cable runs. Well grounded equipment will minimize any RF issues.

    Multiple grounding might actually cause issues if there is stray AC from the high voltage. You can check for stray AC by having some ground rods placed at the site and check for AC potentials between the rods. A single low impedance ground is the best solution, but some electrical contractors don't understand the requirements.

    Also, the cell tower could easily cause interference with WiFi or other RF equipment. If your new plant is going to depend on low power radio IT connectivity for either the plant or for IT, you should have an interference study done.
    • Hrmmm.. We are definitely going to be using WiFi and/or RF equipment in the plant.. Thanks for the information...you (or one of your esteemed colleagues) wouldn't happen to know any texts or websites that go into detail on RF tests, etc.? Unfortunately I know zip about electricity and RF interference...
    • Why go to the trouble of having a formal study conducted when so many Slashdotters here can easily spout off exactly what his needs are, why his design won't work, and why he's an idiot without even knowing a damn thing about his situation.
  • All it means is that the effect decreases at an exponential rate, as opposed to a linear one. Yes, that means the effects will diminish faster, but without knowing the output at the source, quoting the Inverse Square Law doesn't actually tell anybody anything. Knowing that my car can brake from 100 kph to 0 in 10 m doesn't help if I don't know how far away from the wall I am when I slam on the brakes.

    So please, yes I'm aware that the effects diminish rapidly...but that doesn't really answer my question.

  • CRT screen problems (Score:2, Interesting)

    by wimbor (302967)
    Our previous building was located just underneath high-voltage power lines. We had to buy laptops and later stand-alone LCD screens because all CRT screens were wobbling due to interference from the power lines...

    I'm no electrical expert, so I have no idea how far this effect reaches, but I would be very prudent and try to measure/test the area for interference.
    • by ldspartan (14035)
      That'd be electromagnetism at work. Farmer's occasionally steal power by setting up large coils of wire under a high tension transmission line and hooking them up to whatever they wanted power.

      More than anything else, this sounds like a good excuse to get some nice LCDs :).

      --
      lds
      • Farmer's occasionally steal power by setting up large coils of wire under a high tension transmission line and hooking them up to whatever they wanted power.

        It's been a while since I did any physics, but is "stealing" really the correct description for "utilising natural EM radiation" ?

        • Well, you are increasing the load on the power lines, and they're losing more power than they would be otherwise. I remember this coming up in some highschool physics class; I think the power companies prosecuted the farmers :).

          Of course, it could all be an urban legend.

          --
          lds
      • My office at work has just been re-decorated, including trunking on the walls for power and ethernet, and all the monitors along one wall have started wobbling.
        After making a fuss, i've been given a 17" LCD - and very nice it is too - but i'd rather have my old CRT back and not have to sit in such a strong EM field
  • by maggard (5579) <michael@michaelmaggard.com> on Wednesday May 26, 2004 @04:01PM (#9261793) Homepage Journal
    Actually, there are some issues to be considered.

    Inverse-square law not withstanding you're gonna have more noise floating around then other locations will. This won't be so much an issue in the server room as much as on long cable runs, or runs at just the right (wrong) angle. And yes, as irrationial as they may be folks likes and dislikes need to be taken into account.

    Advice?

    1. Make sure your server room is getting clean power. Same for all of your closets. As this is a new facility that should've been a no-brainer by the electrical engineers but double check on the planning and then see to it that there are tests made to confirm. The lines and tower are probably swamped out by the junk coming off the manufacturing equipment but you don't want any of that, whatever the source, giving you gremlins.

    2. Likewise make sure all of your grounds are good and everything is properly shielded. That doesn't mean putting screening over every surface, just regular testing of electrical outlets and network drops. Invest in the proper test equipment and put this on your maintenance schedule, prioritize it in your trouble-shooting procedures, doing so should keep on top of any potential problems.

    3. Keep an eye on your neighbors. Seriously, if something goes wonky for them it might spill over to you. Have a standing order with the security crew that all activity regarding the tower & lines is to be noted and reports copied to IS. That way you folks can correlate any problems.

    4. As others have noted, wireless might take a hit. Before doing any mission critical wireless planning wait until the buildings are done, all of the equipment installed, and everything running. Only once the plant is in regular operation is it worthwhile to get a competent site survey done for deploying wireless. (Note: "Mission critical": The access points for the conference rooms and mahogany row are conveniences where iffy behavior is acceptable. Go ahead and roll those out as part of the move-in, they'll provide you with valuable environmental data should you later want to commit to wireless technologies.)

    5. However unreasonable its true most folks don't like being near high voltage lines. It could be the scale, the humming, suspicion of ill effects, it doesn't matter don't try and fight it. Don't bother putting the picnic tables near the lines, don't ask folks to park near 'em if you can avoid it, etc. Indeed the side of the property closest to the tower and lines is probably the best place to put tractor trailers, storage areas, and other unpopular activities.

    Yeah, your cellphones on average put out more energy into your local environment then the tower and lines do. However the tower and lines are there 24/7, they're big and ultimately pretty powerful, and strange interactions do sometimes occur. Its legitimate to undertake a little proactive planning and just make extra sure electrical systems are properly configured, cable segments are well shielded and grounded, etc.

    Lose sleep over any of this? No. Make sure you do everything right? Extra-yes.

    • Remember that the distance to the nearest point of concern is an estimated 150 meters, also known as "almost two football fields."

      I agree with points 1, 2, and 3 - regardless of the circumstances. Just good operating protocol. As for being concerned about them, bingo there. I would be willing to bet that the 220v lines in the wall powering the air conditioning gives off more EMI (as measured in the server room) than the high power lines 150m away.
  • Free power! (Score:3, Funny)

    by GoRK (10018) <johnl@bPLANCKlurbco.com minus physicist> on Wednesday May 26, 2004 @04:10PM (#9261877) Homepage Journal
    Build youself a hugeass induction coil and suck the power right out of the air! Free electricity for your servers!
  • by Gustavo (64413) on Wednesday May 26, 2004 @04:11PM (#9261888) Homepage
    The book The Practice of System and Network Administration [everythingsysadmin.com] has a few excelent chapters covering everything you need to think about when planning a data center. It's a great book.
  • by ElectricRook (264648) on Wednesday May 26, 2004 @04:19PM (#9261948)
    Remember that power and cooling requirements per rack slot will double every eighteen months.

    That means that when you replace the out dated equipment, the new servers will require more power and cooling than the older equipment. Follow Moore's law (double X every 18 months) and you can't go wrong.

    I only know this, as it what happened to us.

    P.S. The weight will increase too.

  • We have one sprinkler directly on top of the perforated server cage in the smallish computer room. The whole office's sprinklers are interconnected, so if theres a fire in the kitchen,we're down till IBM brings us new machines. If our disks go bad too, I'll have to start browsing dice.com again.

    Considering one full days downtime will cost us an estimated 80,000 $$$, the sprinklers have been high on my priority list to push for change.

    I dont think other companies on our street are in less hazard. Someone
    • Considering one full days downtime will cost us an estimated 80,000 $$$, the sprinklers have been high on my priority list to push for change

      Umm.. That is a no brainer... Get your GM show him the sprinkler in the server room and see.. "We need money to make it go away".. if he doesn't understand ask for his cell phone and ask him to follow you to the bathroom .. He should ask quite quickly what your doing.. then tell him your going to show him what happends when electronics get wet.. he should get the hint
  • I've got a Synchrotron [vic.gov.au] on the other side of the road from my new data center. I know I'll be wishing I had put more shielding in place at some point although I don't know how noisy its going to be.

    One thing to consider for small data centers (mine holds 3 racks) is that modern building construction assumes that the walls go where they build them and the fancy floor gets adjusted to the walls. That turns out to be very nasty when your in a room with 4x7 computer room tiles that are exactly the same size a
  • "Are there any special considerations when designing my server room that I need to be aware of?"
    Um, yes. If, say, your servers are maintained by a third party, they'll have all sorts of environmental requirements (security, hookup to BMS fire alarm, aircon of sufficient quality/performance/with a good maintenance contract) etc etc. Likewise, if your business is insured against the costs (both material and in terms of lost productivity) of damage, then the insurers will want to know you're not keeping the
    • That wasn't really my question; I know the environmental requirements of a server room. They fall just short of talking about cell phone towers and high voltage lines. Someone who specializes in this work will be designing and building the server room. I just want to make sure that I know what I need to ask the vendors, because when we find out the cell phone tower prevents our RF Units from functioning, I'm the one who is going to get in trouble, not the vendor.
      • if you've got standard equipment, it's going to make one metric F-all's worth of difference. If you're using any RF transmitter/receiver equipment (we use ECG telemetry, for example) then your vendor will advise and certify.
        Remember the inverse square law.

Is a person who blows up banks an econoclast?

Working...