Re: Cache traffic - odd?

From: Simon Rainey <>
Date: Thu, 04 Sep 1997 15:10:37 +0100

We run ten heavily used squid caches. These are located at three locations.
Each site has a 100Mbit/s backbone and the servers hang off Cisco Catalyst
10/100 switches.

We use MRTG to monitor the traffic levels for each server, taking the data
from the Catalyst switches.

An analysis of the squid log files shows about 22% less data fetched than
served. However the MRTG traffic analysis shows about equal traffic flow,
and sometimes the inbound traffic greatly exceeds the outbound. Some of
this can be explained by client HTTP requests, sibling UDP queries and DNS
traffic, but I would have thought that these would be small compared to the
actual data served.

I don't doubt for a minute the integrity of the squid logs. I'm more
inclined to mistrust the data I get back from the switches. But then one
might expect any errors affecting the analysis of traffic in one direction
to affect the analysis of traffic in the other direction to an equal
degree. The delta would then be unaffected.

Anyone got any thoughts on this?


At 14:49 04/09/97 +0100, Leigh Porter wrote:
>Hi folks,
>We run two pretty well used cache servers and I have noticed
>some odd traffic stats.
>The idea of a cahce server is to serve objects locally, an offset of
>this is that it saves bandwidth.
>We use MRTG to monitor the ethernet line going to the cache server
>from a Cisco switch and even though the server gets almost a 30%
>HIT rate there is often more traffic going OUT TO THE SERVER than
>there is come FROM THE SERVER - which is odd!
>Occasionally it will go the other way around.
>Anybody know why this is or have similar monitors of bandwidth
>usage they can share!?

Simon Rainey Direct Line: 01235 823238
Principal Internet Development Engineer Fax: 01235 823424
RM Internet for Learning E-mail:
New Mill House, 183 Milton Park, Abingdon, Oxfordshire, OX14 4SE, England
Received on Thu Sep 04 1997 - 07:12:23 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:36:55 MST