Noise Cancellation Of Server Noise

Home » CentOS » Noise Cancellation Of Server Noise
CentOS 26 Comments

Has anyone come across a device that could cancel out the noise from servers, ie. the fan noise? I have a server rack near my office and would like to see if i can decrease the noise level. It would seem to me that the steady drone of the fans could be cancelled out to a large extent.

26 thoughts on - Noise Cancellation Of Server Noise

  • thats really hard to do because the noise source is diffused, and reaches your ears via multiple paths, plus your ears aren’t at a fixed location.

    Its easier to move the servers to somewhere you won’t hear them, like a server closet that you can soundproof and add forced ventilation to.
    or wear earplugs.

  • wow, $9100 for the 38U rack. thats pricey… and that’s before PDUs, UPS, and all that other important rack mount stuff.

  • To be fair, OP didn’t specify a budget. Also, I typo’ed ‘CS’ should have been ‘CX’.

    You can probably DIY a quieter rack for a lot less. There might also be other companies with similar products for a lot less. I don’t know, I
    just remembered seeing the CX in the past.

  • }}

    temperature regulated fan speed can reduce noise.

    use of a baffle box with styrofoam or thin foam rubber lining.

    —–+—–+
    fan \/ | |
    /\ | |
    –+—–+-

  • he said ‘server rack’. thats a probably 6′ tall cabinet filled with servers, each of which has likely dozens of high speed fans.

    (back of 2 server cabinets I used to run at work…)
    https://photos.smugmug.com/By-Date/2015/2015-01-31/n-vn7hn/i-QFKg46V/0/X3/i-QFKg46V-X3.jpg

    (front of one of those cabinets, https://photos.smugmug.com/By-Date/2015/2015-01-31/n-vn7hn/i-cJs8Dkk/1/X3/i-cJs8Dkk-X3.jpg

    a cabinet like that needs 600 cubic feet/minute or more airflow.

    now, I’m guessing that H’s rack isn’t quite this big, or as densely populated, since its not in a dedicated server room, but still, it has a lot more than a couple fans to silence. my rack there has 8 servers, 2
    additional disk trays, and about 200 disk drives, also a pair of 7000 VA
    UPS’s in the bottom (actual load was around 3000VA)

  • <>

    }}

    true. i was thinking ‘old school’ slide drawer racks.

    because i doubt that ‘it’ or ‘heads of authority’ will allow changes to server racks, H’s best bet is to get ‘fixed base’ or ‘roll around’ type acoustic wall panels. they are available with various types of sound absorbing covering and of various heights and widths.

    one or two can be placed between him and area of rack to block/absorb sound.

  • If the server decision makers had not gone for a server designed by accountants but a server designed by engineers then you would not have this problem.

    Regards, Mark Woolfson MW Consultancy Ltd Leeds United Kingdom Tel: +44 113 259 0759
    Mob: +44 786 065 2778
    —–Original Message—

  • }}

    maybe accountants in U.K., not is U.S.A..

    accountants in U.S.A. only ‘know’ about spreadsheets because that is what they are paid for.

    engineering is by engineers because that is what they are paid for.

    neither is going to do something they are not paid for.

    long live the “American Way”.

    ((GBWG))

  • Lack of vision and inspiration always hampers the betterment of our unique world. Great people, all over our planet, usually do a bit more than they are paid for. The resulting benefit for others is incalculable.

    Just think of what the CentOS founders did, unpaid, and their hardworking successes achieved – never an eye on the clock and never restricting their efforts to make something not just work, but work well.

  • H wrote on 24/12/2016 22:43:

    It’s probably too late for you, but the best solution is to buy office-grade servers instead of datacentre-grade servers

    HP and Dell (don’t know othjer brands so well) have two kinds of servers:
    “dense” rack-only servers (tipically 1U or 2U if we are talking about single servers and not Apollo or VRTX or blade) and “office type” servers, 4U to 6U.

    Let’s put aside Apollo, VRTX, and other blade, they are out of scope.

    Dnese servers are extremely noisy because they have a lot of small fans turning relatively at high speed. This kind of server cannot be converted to free standing “tower” form factor.

    The other kind of server is sold both in “tower” and rack form factor; they are bigger the airflow is forced by bigger fans that can turn to a lower speed to exchange the heat.

    If the noise is produced by the rack fans, use, as someone else suggested, variable speed fans regulated by a temperature sensor.

    In essence, you can greatly lower the noise “by design”, but if you already have the equipment it can be too late.

  • }}

    this is true,

    }}

    ‘great people’ are usually the exception.

    }}

    other exception as near all things connected FSF and Linux.

    both are ‘given’ “in a world with out fences”.

  • }}

    fine looking box. oem, model?

    would like above to pull manuals.

    awaict, 6 fans, 4 power style connectors = power+tack, or sata drive, or 3 volt supply connectors.

    }}

    note:: easy

    male to female fan power ‘dongle’ with temp to speed controller.
    insert between fan power cable and fan power socket.

    do not claim to be ‘it’. nor do i came dcs, dce.

    do claim;
    design ic control systems – 1969,
    build, maintain, service, s100 micro computers – 1974

    so, yes, easy… ;=)

    if you run a web search for such a ‘dongle’ available, you should find plenty.

    rotron.com is down, can not supply url.

  • geo.inbox.ignored wrote on 26/12/2016 07:22:

    1U and 2U HP/Dell servers use variable speed fans, you hear them at full speed when you power the server on.

    The problem is that the noise of an “idle” 1U or 2U server is far greater than the noise of a 5U server for three main reasons.

    1. The small servers have less airflow space, so if you want to move (say) 1
    cubic meter of air per hour you have less space, so you must increase the speed of the air

    2. The fans are smaller, so the blades must turn faster to move the same amount of air

    3, Small servers have smaller dissipating area/volume, so to exchange the same amount of heat you must increase the airflow speed.

  • The worst part in the approach “make rackmount machines quieter” is: these machines are designed with high density of power per physical space and efficient cooling in mind (as computers do not produce mechanical work, hence virtually all consumed power goes into heat). Noise level is never considered as a goal. So, attempting to make rackmount machine quieter is counter productive.

    Just my $0.02

    Valeri

    ++++++++++++++++++++++++++++++++++++++++
    Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247
    ++++++++++++++++++++++++++++++++++++++++

  • I personally do not consider these tower machines put sideways rackmount servers. Such tower machine is usually not designed to works in the server room (where cost per square foot of floor space per Month is consideration), a bunch of machines like that can be replaced by single real rackmount server, the last usually would be much more robust and durable as they are designed to work more stressed. Of course, server room full of tower machines would be quieter, but that one I would rather describe as room full of servers instead of server room (should use the word desktops or workstations not servers for the former).

    I should have added “rant” tags… I have mild excuse though: I do have my share of trouble from these put on the side towers (chosen by somebody else) in my server room.

    Valeri

    ++++++++++++++++++++++++++++++++++++++++
    Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247
    ++++++++++++++++++++++++++++++++++++++++

  • There are other great systems the above will be true about. Also, many free software developers are actually paid for their software development work. Maybe not as much as one might be paid for commercial software. Great example is postfix (free open source software), architectured, written and developed under the hood of IBM (thank you, great guys!) by brilliant person Vietse Venema (thank you, Vietse!!).

    Valeri

    ++++++++++++++++++++++++++++++++++++++++
    Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247
    ++++++++++++++++++++++++++++++++++++++++

  • accountants but a server designed by engineers then you would not have this problem. unique world. restricting their efforts to make something not just work, but work well. free software developers are actually paid for their software development work. Maybe not as much as one might be paid for commercial software. Great example is postfix (free open source software), architectured, written and developed under the hood of IBM (thank you, great guys!) by brilliant person Vietse Venema

    Wietse Venema

    (I don’t know where to hide in shame)

    Valeri

    (thank you, Vietse!!).

    ++++++++++++++++++++++++++++++++++++++++
    Valeri Galtsev Sr System Administrator Department of Astronomy and Astrophysics Kavli Institute for Cosmological Physics University of Chicago Phone: 773-702-4247
    ++++++++++++++++++++++++++++++++++++++++

  • Interesting. In my case i wanted to decrease the amount of noise escaping through an open door (which can not be closed.) After reading the various comments i think I might search for acoustic blankets for musicians.

  • thats a SuperMicro SC846 storage server chassis (36 x 3.5″ SAS drives in a 4U) with a dual xeon x5650 motherboard. I just used the picture cuz I had it handy and its typical of a rack mount server, big array of fans in the middle, pulling air past the disks and through the mainboard heat sinks via that clear air duct