[c-nsp] Cisco 3550 counters and QoS

[Marco Matarazzo]

> Search back a few weeks/months, but there was just a thread on 3550
> policing and how for reasons nobody could figure out (including the tac
> case I opened) matching on ip dscp 0 will not always police ingress
> traffic.  When I originally tested it in the lab, it worked, but I found
> in production it did not.

This is why I'm testing the device. For now, and for the applications my
customers use, it seems working. We'll see! ;)

> > If I upload anything from the customer machine, the traffic gets policed
at
> > 1Mbps, but on the graphs, I see on the ingress of the layer3 interface
> > 1.20Mbps, and on the egress of the trunk the nice flat line at 1Mbps.
> > Checking the ftp client, I can see it's uploading at 125KBps (that's
> > 1Mbps!), checking the "sh int":
>
> My guess is that the 1.2mbps you're seeing is the policed system trying to
> send slightly more than it's allowed.  The extra .2mbps is being dropped
> by the switch.  Since that traffic is getting to the switch, it doesn't
> seem unreasonable for the snmp counters to include it.  It sounds like
> what you want is an additional per-interface snmp counter for traffic
> actually switched/forwarded.

I think the situation is more or less this one:

Packets get to the switch port, and are counted. Then the policer gets in,
so they're not delivered. So basically what I'm getting is number of packets
hitting the switching engine, not packets that are actually routed
somewhere. I'll try to play with the QOS Snmp counters to have a better
picture...

Thanks!
]\/[arco