We have sensors in our datacenters that report, among other things, humidity. One data center is exceeding the predefined humidity alerts of the devices, but given that cisco says the operating humidty of routers is 10% to 85%, I dont know if humidity of, say, 65% is anything to worry about. Anyone know if increased humidity correlates with decreased MTBF? And reasonable thresholds that I can tolerate, and when I should complain to the datacenter operators? (References I can point to would be nice.) Thanks
| and when I should | complain to the datacenter operators? (References I can point to would | be nice.) When your equipment starts to rust ;) I don't have any technical references, but I think that anything over 65% is probably too much. Most facilities I have equipment in do not exceed that mark. Todd --
IIRC, too low a humidity level makes static electricity a problem. Too high makes the cold air condense on your equipment. 60-65% sounds about right. Todd Mitchell - lists wrote:
| and when I should | complain to the datacenter operators? (References I can point to would | be nice.)
When your equipment starts to rust ;)
I don't have any technical references, but I think that anything over 65% is probably too much. Most facilities I have equipment in do not exceed that mark.
Todd
--
Todd Mitchell - lists wrote:
| and when I should | complain to the datacenter operators? (References I can point to would | be nice.)
When your equipment starts to rust ;)
I don't have any technical references, but I think that anything over 65% is probably too much. Most facilities I have equipment in do not exceed that mark.
It isn't a *huge* issue (within reason) unless you have printers around and the inherent paper. Big lasers are notoriously finicky about humidity, which directly affects the paper quality. Otherwise, just keep it well below the dewpoint :-) Jeff
On Mon, 04 Aug 2003 23:35:24 -0400 Jeff Kell <jeff-kell@utc.edu> wrote:
Todd Mitchell - lists wrote:
| and when I should | complain to the datacenter operators? (References I can point to would | be nice.)
When your equipment starts to rust ;)
I don't have any technical references, but I think that anything over 65% is probably too much. Most facilities I have equipment in do not exceed that mark.
It isn't a *huge* issue (within reason) unless you have printers around and the inherent paper. Big lasers are notoriously finicky about humidity, which directly affects the paper quality.
If you are running lots of mag tape, humidity > 60 % starts to really increase tape and head wear. If new tape heads are part of your regular operating budget, I would keep the humidity < 50 %. Regards Marshall Eubanks
Otherwise, just keep it well below the dewpoint :-)
Jeff
On Mon, Aug 04, 2003 at 01:02:46PM -0700, Steve Francis wrote:
And reasonable thresholds that I can tolerate, and when I should complain to the datacenter operators? (References I can point to would be nice.)
depends what's in your SLA. If it states 40-65%, and you notice it's over 68% or so, you SHOULD complain. Otherwise, you "agree" to the higher %, if they can prove you know about the higher %. (IANAL)... Of course, there's a difference between complaining, and demanding refund. However, I think it's wise to complain... Kind Regards, Frank Louwers -- Openminds bvba www.openminds.be Tweebruggenstraat 16 - 9000 Gent - Belgium
For a datacenter with a single controlled area the ideal range for relative humidity (RH) is in the neighborhood of 35% to 50%. Here are some data points: 1) Static electricity is minimized when RH is at or above 35%. 2) RH below 25% can cause embrittlement of hygroscopic materials such as paper. 3) RH above 65% can cause mold growth and metal corrosion. 4) Humans are most comfortable when the RH is between 20% and 60%. 5) RH above 50% in cold weather can cause condensation inside of outer walls (or on equipment itself if the facility has an external door or window that opens directly into the controlled area). Sean Donelan wrote an informative post on a related subject a few year back: <http://www.cctec.com/maillists/nanog/historical/9810/msg00276.html> -Richard
participants (7)
-
Frank Louwers
-
Jeff Kell
-
Marshall Eubanks
-
N. Richard Solis
-
Richard Parker
-
Steve Francis
-
Todd Mitchell - lists