04-11-2006 01:29 PM - edited 03-17-2019 08:40 PM
I am wokring on prioritizing our Voice traffic over our 10mb fiber link between our office and our hub, that houses our PRI gateway. Here is the config that i pu ton both sides of the link. I still see random call problems wich are usually voice dropping so that words here and there get missed.
class-map match-any VoIP-Control-UnTrust
match access-group name VoIP-Control
class-map match-any VoIP-RTP-UnTrust
match protocol rtp audio
match access-group name VoIP-RTCP
class-map match-any VoIP-Remark
match ip dscp ef
match ip dscp cs3
match ip dscp af31
!
!
policy-map Policy-UnTrust
class VoIP-RTP-UnTrust
priority percent 70
set dscp ef
class VoIP-Control-UnTrust
bandwidth percent 5
set dscp af31
class VoIP-Remark
set dscp default
class class-default
fair-queue
!
interface FastEthernet0/0
description Fiber to Hub
no ip address
duplex full
speed 10
service-policy output Policy-UnTrust
!
interface FastEthernet0/0.7
description Fiber to Hub
encapsulation dot1Q 7
ip address 7.x.x.x.255.255.248
no ip redirects
no ip mroute-cache
no snmp trap link-status
no cdp enable
!
ip access-list extended VoIP-Control
permit tcp any any eq 1720
permit tcp any any range 11000 11999
permit udp any any eq 2427
permit tcp any any eq 2428
permit tcp any any range 2000 2002
permit udp any any eq 1719
permit udp any any eq 5060
ip access-list extended VoIP-RTCP
permit udp any any range 16384 32767
This is on both sides of the link, ans i am still seeing problems, that seem to be related to other traffic taking bandwidth from the voice.
04-12-2006 08:05 PM
It is unusual to mark packets and set up your queues with the same policy map, but having never tried it I can't say it won't work. If the voice endpoints are Cisco gateways or phones the DSCP values should already be set correctly, in which case you can just define your class maps to match ip dscp or ip precedence. If not Cisco, you might try using a policy map applied as an input service policy on the ingress interface to mark packets. Then your egress policy could just set up the bandwidth for your media and control traffic.
Have you done a show policy-map interface f0/0? This will show you what your service policy is actually doing. It would be especially useful if you can look when the quality is bad.
Do you have CAC to prevent the voice traffic from exceeding the size of the priority queue? Anything exceeding your 70% will be dropped even if bandwidth is available.
One thing that can cause clipping is having VAD enabled. Are your problems only happening when the link is heavily loaded? If all the time look at VAD.
Please rate helpful posts.
04-13-2006 07:41 AM
Thanks for all your help. First i will admit that QOS is totally new to me and I am strugling.(taking course real soon)
All of my phones (non-cisco) are on a specific range of IP's. 10.7.7.20 to 10.7.7.50 Can I, on the ingress interface assign them a DSCP value that is matched on the outgoing interface that gives them high priority? Is this what you were saying to me? Is this a good idea?
Thanks again.
04-13-2006 10:22 AM
As always the best solution depends on a lot of factors. In general, it is best practice to mark packets as close to the sending device as possible. If you have a Cisco switch you may be able to mark packets there which offers the best performance. Even though your phones are non-Cisco I bet they are capable of setting DSCP. This would simplify things, as you would just need to trust those settings, and then use them to setup your egress QOS. Is the gateway Cisco?
To answer your specific question, yes that would work , though it is important to differentiate between media and control traffic, so you will need to pin it down to ports as well as IP.
04-13-2006 12:27 PM
Upon a bit more investgation I noticed that the phones send all of thier data to our voice server (10.7.7.19) here at this office, and then it sends the traffic to our voice server (7.7.1.19) in our hub, which in turn sends it to the Cisco voice gateway (10.254.254.3). At no time does the voice go directly to the GW or to the hub voice server (10.7.7.19).
So here is what I have and it seems to be helping.
class-map match-any voip-priority
match ip dscp ef
class-map match-any voip-tag
match access-group name voip-tag
!
!
policy-map voip-priority
class voip-priority
bandwidth 7500
class class-default
fair-queue
queue-limit 5
policy-map voip-traffic
class voip-tag
set ip dscp ef
class class-default
set ip dscp default
!
interface FastEthernet0/0
description Fiber to hub
no ip address
duplex full
speed 10
service-policy output voip-priority
!
interface FastEthernet0/0.7
encapsulation dot1Q 7
ip address 7.7.1.1 255.255.255.0
no ip redirects
no ip mroute-cache
no snmp trap link-status
no cdp enable
!
interface FastEthernet0/1
description Hub to Office
ip address 10.7.7.9 255.255.255.0
duplex full
speed 100
service-policy input voip-traffic
!
ip access-list extended voip-tag
permit ip host 10.7.7.19 host 7.7.1.19
!
I should have this similarly setup on the other side of my link right? that way it prioritises the traffic in both directions, right?
04-14-2006 07:20 AM
Yes, you need to set up both directions. Does a show policy-map int FX/X verify that packets are being marked on F0/1 and priority queued on F0/0?
Please the use the pulldown menu to rate helpful posts.
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: