RE: [squid-users] Strange performance effects on squid during off peak hours

> ons 2010-09-15 klockan 14:01 +0100 skrev Martin Sperl:
> > Essentially we see that during peak hours the Average response time is
> better than during off-peak hours.
>
> Average response time measured on what?
Measured via the access log of squid.

>
> Every request being handled, or the response time of some well known
> synthetic request?
All requests show this statistical behavior
Also our "synthetic tests" that are measured externally every 5 minutes.

>
> If measured over all requests then it may well be normal. Under low
> traffic conditions a couple slow requests such as downloads to a low
> bandwidth client has big impact on the statistics, while under high
> traffic those drowns in the rest of the traffic.
>
> I am afraid you need to drill down a bit in the data to tell what this
> really is about. May be perfectly normal or may be a sign of problems.
> Can't tell from the statistics alone.

I know, that is why I have created and added the histogram data, that shows that say at 6am UTC >90% of the requests are above 0.03 seconds. While during peak hours (say 6pm) we have a peak at 0.011 seconds and only <5% above 0.030 seconds

As Amos has said: if there are not enough "requests", the event driven design may be the "culprit" for the introduced latencies...

I will need to investigate this...

Thanks,
        Martin



This message and the information contained herein is proprietary and confidential and subject to the Amdocs policy statement,
you may review at http://www.amdocs.com/email_disclaimer.asp
Received on Wed Sep 15 2010 - 19:30:15 MDT