Do I need to do QoS on a voice gateway?
I have a cisco router 2951 using it as a voice gateway. Currently we are using 2 PRI to get to the PSTN and a gig port connected to the core 4507.
My question is, do need to mark the traffic coming from the PRI or is it already mark. Where do we mark it if it need to be mark. Or, Do we even need to mark it at all at the voice gateway or just mark it at the incoming interface at the 4507?
You didn't mention if it was configured as SIP, H323 or MGCP.
For your voip dial peers you should at least mark CS3 for signaling. EF should be default.
Please rate if this helps.
You can always use Wireshark to verify the markings out the interface.
Then you can always use an extended ping and set the TOS you would like to mark. EF is 184.
Typically you can set in on the voip dial peer as such if needed:
ip qos dscp ef media ip qos dscp cs3 signaling
In this page http://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/collab10/collab10/netstruc.html#pgfId-1044051
What is the difference between Layer-2 Classification and layer 3 Layer-3 Classification ?
Where should we use Layer-2 Classificationand Layer-3 Classification ?
L2 classification (CoS) is appended in dot1q header while L3 classification (IPP or DSCP) is appended in IP header.
I prefer to use L3 all the way and that's the preference in Cisco QoS SRND. The reason that if you rely on CoS matching then the packet traverse L3 link or L2 access port, CoS marking will be stripped as there won't be dot1q header. At that time you need to use DSCP which means that you need to make sure that you always have consistent mapping between DSCP and CoS across the network. If you use DSCP all the time end-to-end then you are running homogeneous QoS classification/marking and less error approach. Similarly, if you cross to 3rd party network (e.g. MPLS provider) they don't consider CoS. Instead they use DSCP to classified and insert request MPLS labels for traffic engineering.
*** Please remember to rate useful posts
You want to mark the traffic as close to the edge as possible.
If you check the SRND:
"It has always been an integral part of the Cisco network design architecture to classify or mark traffic as close to the edge of the network as possible."
In this case it would be your gateway.
The default for signaling is af31, as Jason suggested you would like to change this to cs3
For H323 GW
dial-peer voice 1000 voip
ip qos dscp cs3 signaling
For MGCP GW
mgcp ip qos dscp cs3 signaling
Verify the settings with:
show dial-peer voice 1000 | include DSCP
I have a similar question but in my case, the router (PSTN gateway) is configured as a SIP trunk to CUCM (it's SIP between CUCM and the router/PSTN gateway, but the actual PSTN access is ISDN PRI). What would be the appropriate settings on the router used as the PSTN gateway in my situation? I assume it's just an entry on the voip dial-peers? If so, what would be the correct entry? And would the settings be the same when we finally replace our ISDN trunks with SIP trunks to the phone company?
I saw this posting: https://supportforums.cisco.com/discussion/10095976/ip-qos-dscp-ef-media
So I assume the voip dial-peer is already marked as EF for the voice traffic by default unless you intentionally change it to something else, correct? If it's EF by default, then that's great because that's what I need.
When I look at one of our voip dial-peers (show dial-peer voice 1), I see this, which does look like it's already marked as EF...
ip media DSCP = ef, ip media rsvp-pass DSCP = ef
ip media rsvp-fail DSCP = ef, ip signaling DSCP = af31,
ip video rsvp-none DSCP = af41,ip video rsvp-pass DSCP = af41
ip video rsvp-fail DSCP = af41,