[c-nsp] High drop rate with not much utilization

[Rodney Dunn]

It's almost always microburst from the LAN going over the WAN side BW.

Try this, remove the egress QOS policy and put a simple rate limiter on 
there to conform at 40 Mbps and for exceeds just transmit.

Then watch to see if you see exceeds. If so you are overrunning the 
egress side BW and the buffers are not deep enough to handle it.

That's the most common when the LAN side is a higher speed..ie: Gige.

Rodney



On 4/8/10 5:57 PM, Sergio D. wrote:
> Hello,
> We currently have a DS3 with an outbound service policy prioritizing voice
> traffic, some of this voice is not marked correctly (a different issue we
> are working on) and it ends up on the class-default queue. The problem is
> that we drop packets from that class-default queue even if our outbound
> utilization is less than %50. Can someone explain why that happens? I
> understand that it could possibly be top secret Cisco algorithm stuff.
> Here is a snapshot of when it happens:
> Cisco 3845 btw.
>
>   1
>   2
>   3
>   4
>   5
>   6
>   7
>   8
>   9
> 10
> 11
> 12
> 13
> 14
> 15
>
> 3845-UT02#sh policy-map interface s1/0 output class class-default
>   Serial1/0
>
>    Service-policy output: Qos-Policy
>
>      Class-map: class-default (match-any)
>        1867364 packets, 271186963 bytes
>        5 minute offered rate 13938000 bps, drop rate 15000 bps
>        Match: any
>        Queueing
>          Flow Based Fair Queueing
>          Maximum Number of Hashed Queues 256
>          (total queued/total drops/no-buffer drops) 0/984/03845-UT02#sh
> int s1/0 | in tx|rx
>       reliability 255/255, txload 86/255, rxload 91/255
>
>
>
> Thanks,
> Sergio Danelli
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/