cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
618
Views
0
Helpful
5
Replies

QoS for PABX Voice Traffic

connie
Level 1
Level 1

Hi All

We have two sites - each with its own PABX, and a 320KB direct line between them. This line is being used for data, but we want to route PABX voice traffic over it as well.

This is where the problem comes in. I went to http://www.cisco.com/en/US/customer/tech/tk652/tk698/technologies_white_paper09186a00800d6b73.shtml#xtocid11 to have a look at what Cisco recommends. I configured both the 2620 routers linking the sites as described in the chapter http://www.cisco.com/en/US/customer/tech/tk652/tk698/technologies_white_paper09186a00800d6b73.shtml#69895.

This, however, did not work too well. :-(

The two PABXs have fixed IP addresses, so identifying and tagging the traffic that needs to be shaped is quite easy (using an ACL). After this solution had been implemented, general performance over this link became so bad, that we had to revert back to the original config to restore normal service.

The problem with our voice, is that it starts breaking up quote badly as soon as we reach 80K on our 320K link. Hence the QoS attempt.

Does anybody have any thoughts? I upgraded both routers IOS's to 12.2(29).

5 Replies 5

dsouzavincent
Level 1
Level 1

Hi,

Cud you attached the config's you tried. and the original config of the router before the change.

Regards,

Vincent

Hi Vincent

The config on both routers look like this (except for the IP addresses being different, of course). I have removed some of the more sensitive information, but take my word for it that connectivity is not an issue. ;-)

We are using interface s0/1 on both routers for this 320K link. The PABX IP address on site A is 10.30.200.1, and on site B is is 10.34.200.1.

Config before:

---------------------------

version 12.2

service timestamps debug uptime

service timestamps log datetime

service password-encryption

!

hostname

!

boot system flash:c2600-js-mz.122-29.bin

logging buffered 20000 debugging

aaa new-model

aaa authentication login vtymethod group radius enable

enable secret 5

!

ip subnet-zero

no ip rcmd domain-lookup

ip rcmd rsh-enable

!

!

interface FastEthernet0/0

ip address 10.32.5.253 255.255.255.0

speed 100

full-duplex

!

interface Serial0/0

bandwidth 192

ip address

no ip mroute-cache

!

interface Serial0/1

bandwidth 320

ip address

no ip mroute-cache

!

router eigrp 2050

variance 2

redistribute static

no auto-summary

no eigrp log-neighbor-changes

!

ip classless

ip route 0.0.0.0 0.0.0.0 10.1.1.254

no ip http server

!

!

snmp-server engineID local xxx

snmp-server community RO

snmp-server community RW

snmp-server enable traps tty

radius-server host 10.1.1.201 auth-port 1645 acct-port 1646

radius-server key 7

!

line con 0

exec-timeout 0 0

line aux 0

line vty 0 4

exec-timeout 10 10

password 7

logging synchronous

login authentication vtymethod

length 14

width 162

!

end

Config After:

---------------------------

version 12.2

service timestamps debug uptime

service timestamps log datetime

service password-encryption

!

hostname Hurteria2620

!

boot system flash:c2600-js-mz.122-29.bin

logging buffered 20000 debugging

aaa new-model

aaa authentication login vtymethod group radius enable

enable secret 5

!

ip subnet-zero

no ip rcmd domain-lookup

ip rcmd rsh-enable

!

!

class-map match-all C_VoIP

match access-group name A_VoIP

match ip precedence 5

!

!

policy-map P_VoIP

class C_VoIP

priority 64

class class-default

fair-queue

!

!

!

interface Multilink1

bandwidth 320

ip address

service-policy output P_VoIP

no ip mroute-cache

no cdp enable

ip rtp header-compression iphc-format

ip tcp header-compression iphc-format

ppp multilink

ppp multilink fragment-delay 10

ppp multilink interleave

multilink-group 1

!

interface FastEthernet0/0

ip address 10.32.5.253 255.255.255.0

speed 100

full-duplex

!

interface Serial0/0

bandwidth 192

ip address

no ip mroute-cache

no fair-queue

!

interface Serial0/1

bandwidth 320

encapsulation ppp

ppp multilink

multilink-group 1

no ip mroute-cache

!

router eigrp 2050

variance 2

redistribute static

no auto-summary

no eigrp log-neighbor-changes

!

ip classless

ip route 0.0.0.0 0.0.0.0 10.1.1.254

no ip http server

!

!

ip access-list extended A_VoIP

permit ip host 10.30.200.1 host 10.34.200.1

permit ip host 10.34.200.1 host 10.30.200.1

snmp-server engineID local xxx

snmp-server community RO

snmp-server community RW

snmp-server enable traps tty

radius-server host 10.1.1.201 auth-port 1645 acct-port 1646

radius-server key 7

!

!

!

line con 0

exec-timeout 0 0

line aux 0

line vty 0 4

exec-timeout 10 10

password 7

logging synchronous

login authentication vtymethod

length 14

width 162

!

end

How much voice traffic are you sending accross, the policy-map P_VoIP

class C_VoIP

priority 64

policy map strickly prioritizes 64kbps of traffic defined by your ACL, if you are using g711 codec over mp with comression that's 67.6 kbps for RTP stream plus additional signaling overhead (couple kbps). With g729 that's 11.6kbps. With LLQ (priority) if excess bandwidth is available, the priority class will be allowed to utilize the bandwidth. If no excess bandwidth is available, the priority traffic will be constrained to the configured rate via packet drops. Make sure 64k is enough bandwidth for that traffic.

Chris

Thanks for the reply Chris.

We have four voice channels available between the two PABX's (Siemens HiPath 300). Each of these channels is configured to use 17K of bandwidth. We had the channels down to 3 at one stage, and we even had the priority bandwidth up to 128K. It made no difference.

After about 1 hour of configuring the routers, we started having problem with both voice and data transmission. On our bandwidth graphs, the link never exceeded 54K!

The routers on site A and B both use interface s0/0 to connect back to our head office. Both sites had trouble connecting back to our head office after about an hour of changing the config.

Initially, the config worked fine. We managed to make voice call while the link was running at 317K. Then quality started degrading until such a stage that neither voice nor data was at acceptable levels. Like I said, 54K was the maximum throughput we got...

IOS bug?

It sounds like your voice is not hitting the priority queue. show policy-map interface mu1 will tell you.

If you are not seeing the voice in your priority queue, I would suggest applying an inbound service policy on your ethernet to set the IP Precedence, then matching IP Precedence on the outbound service policy. The following changes to your config should get this working:

class-map match-all voice

match ip precedence 5

class-map match-all pabx

match access-group name A_VoIP

policy-map Set_IP_Prec

class pabx

set ip precedence 5

policy-map P_VoIP

class voice

priority 64

class class-default

fair-queue

interface fast0/0

service-policy input Set_IP_Prec

interface multilink1

service-policy output P_VoIP

Review Cisco Networking for a $25 gift card