[c-nsp] Safe operating temps - Somewhat OT

[Nick Hilliard]

On 28/02/2010 14:55, Graham Wooden wrote:
> The AC unit in my little datacenter room was recently upgraded/serviced.
> Since then I have started to monitor the my 6500's temp. I have been
> noticing a swing in the RP in/out on a Sup32 going from 37c to 39c and then
> back to 37 within 30-40 minutes timeframe. Room temp seems to be stable
> right at 25c, all the time.
> 
> Are these safe swings? Or are these too dramatic only to shorten the life of
> the box this much sooner. Unfortunately I don't have any previous history
> data on the sups, only the room.

2C shifts should be ok.  Intel did some interesting research on this a
couple of years ago.  You can read / watch a summary of it here:

> http://www.intel.com/it/pdf/Reducing_Data_Center_Cost_with_an_Air_Economizer.pdf
> http://video.intel.com/?fr_story=2d6e0fbbef76b72c6119cc7fe7889bba20cb5192&rf=bm

synopsis: a production server room hosting 450 server was operated for 10
months using outside air cooling measured temperature fluctuations of 18C
to 33C, and humidity from 4% to 90%.  Ambient dust was prevalent.   Server
failure rate over the 10 months was 4.46%, compared 3.83% in the control
configuration (a standard DX air-conditioned server room hosting the same
server configuration).

Nick