[c-nsp] relation between heat and packet-loss

[Martin T]

This is a good point. However, what might cause switch failure in case
of high temperature? I mean I can understand when transistors of the
networkswitch CPU switch by themselves(resistance of semiconductors
should decrease with the increase of temperature which might cause
faulty switches), but I don't believe that (very) high room
temperature could cause such problems.. Or is there some sort of
temperature monitoring built into IOS, which will for example restart
the networkswitch when temperature has reached a certain level? If
yes, then is it possible to view the value of this threshold?

regards,
martin


2011/5/26 Peter Hicks <peter.hicks at poggs.co.uk>:
>
> On 26 May 2011, at 10:54, Martin T wrote:
>
>> I have a 1U server in the data-center, which is connected trough
>> digital distribution frames to ISP Cisco 4500 series switch and from
>> this switch to Cisco 7200 series router. ISP switch and router are in
>> the same room(room A). Server is in another room(room B). Previous
>> weekend I noticed heavy packet loss to my server and when I connected
>> to the server over out-of-band management(another ISP, no equipment in
>> room A) and pinged the default gateway of my server(Cisco 7200 in room
>> A) results were around 90% packet loss. I'm aware, that there was some
>> sort of AC malfunction in room A and that was the reason ISP provided
>> to me in order to explain this heavy packet loss, but how could
>> increase of temperature cause such packet loss?
>
> Maybe look wider - what if one of their switches failed upstream and you and others were going through an alternate switched path with insufficient bandwidth to cater for failover - say, 10Mbps in failover versus 1Gbps in live.
>
>
> Peter
>
>
>