So, I'm keeping an eye on my MPLS router for VTC traffic bound between offices. I came across this link, that describes what some of the stats mean when you issue a "show policy-mapt interface" command.
http://www.cisco.com/en/US/tech/tk543/tk760/technologies_tech_note09186a0080108e2d.shtml
According to this document, if the # of packets is not significantly larger than the number of packets matched, you are filling the TX ring (hardware queue) and packets are then being process or cef switched at L3. If I stated that right, anyway. This can cause performance issues and it's indicative of congestion.
So, I am looking at stats on my router, and I noticed that consistently, there is only ever 1 packet more in the packets count than in the packets matched count. Does that mean my TX-ring can only hold 1 packet? I know there is a command to set a tx-ring-limit but if it's really a hardware queue, I'm not sure how you can use a software command to expand it's capacity and I'm not sure what that would do to other functions.
Below are some stats off my router, a 2851. One thing is that even with congestion I'm not dropping packets so, maybe that's okay. But this is with only one VTC call up and running. If more calls started up, maybe that could change. The question is, should I be worried about the closeness of these to counter values?
2555_2851#show policy-map int mu100
Multilink100
Service-policy output: QoS_POLICY
Class-map: GOLD (match-any)
550692 packets, 372363191 bytes
5 minute offered rate 164000 bps, drop rate 0 bps
Match: access-group 180
550691 packets, 372362086 bytes
5 minute rate 164000 bps
QoS Set
dscp ef
Packets marked 550692
Queueing
Strict Priority
Output Queue: Conversation 264
Bandwidth 4608 (kbps) Burst 115200 (Bytes)
(pkts matched/bytes matched) 24142/4505482
(total drops/bytes drops) 0/0
