[mrtg] Bandwidth of entire Switch

[Steve Shipway]

(disclaimer: I'm biased in favour of routers2 of course)

> From: mrtg-bounces at lists.oetiker.ch
> [mailto:mrtg-bounces at lists.oetiker.ch] On Behalf Of Daniel J McDonald
>
> > Why couldn't he just do something like:
> >
> > Target[totalswitch]: 1:public at 192.168.1.10 + 2:public at 192.168.1.10 +
> > 3:public at 192.168.1.10 + 4:public at 192.168.1.10:
>
> Because the cumulative rollover effect gives vastly unpredictable
> results.  MRTG can cope with a single counter rollover.  Trying to sum
> 52 gigabit interfaces and make sense of the rollovers is sheer
madness.

This is correct, MRTG will not be able to cope with rollover precidion
if the numbers are this big.  Also, you don't have the opportunity to
identify which port is the source of the most traffic.

> I will look into the router2 I have not checked that out yet.

Here is the demo page:
http://www.steveshipway.org/cgi-bin/routers2.pl?rtr=switches%2Fswitch.cf
g&xgtype=w&if=_incoming
This page is not an ideal example (of the three interfaces being
totalled, two have near zero throughput) but it helps.  Also it uses the
old RRD 1.0, so if you have RRD 1.2 then you'll get smoother and
prettier graph images.

If you have a lot of ports on your switch, you enable the 'active'
option for the userdefined graph and it hides any inactive ports on the
graph.

> I think I might also look into what was suggested ealier about simply
> monitoring the uplink ports for an over all for each switch.

This is what we use here as an indication of switch business, although
as we have routers2 we get the port summary as well.

If you have problems with routers2 then there is a support forum on
http://www.steveshipway.org/forum

Steve