VoIP Outbound Dialpeer

cisco_nerd

So i'm back in a pickle with delayed VoIP calls between CME routers...

Situation:
I have multiple CME routers connected through Gigabit networks (7 in total - not my ideal implementation but I had no say in how it was implemented. Prefered a CUCM pub/sub setup but this is what I was told to roll out). Anyway... my latency is low to none - avg. <5ms. My dial-peer config is such that there is no overlaps with destination-patterns (confirmed through testing all numbers with "sh dialplan number <string>"). Outbound calls to PSTN are specifically matched with access codes and the like. All internal numbers are 5 digit extensions.

Issue:
For local company dialling I have 5 digit dialling between sites with VoIP dial-peers, each site is assigned a logical group of numbers so that they do not overlap with another site, allowing me to configure the dial-peers such that there is only one best route pattern for the call. With this setup, calls local to the CME are obviously instant once the 5th digit is matched and call setup is 1 or 2 seconds. The problem arises when I dial between the CME routers where a delay of 10 seconds is imposed on the final outbound leg for the call.

I have done multiple debugging (debug voip dialpeer) of the calls in both directions on two routers with PuTTY side-by-side and found that on the calling router the outbound dial-peer is matched almost instantly and then the inbound dial-peer on the called router is matched at almost the same time. This shows me that my call is getting to the remote end all well and good as initially I thought the issue would lie with QoS on the network and voice traffic being choked. At this point I see the delay, the router has matched my default inbound VoIP dial-peer and then there is a 10 second delay before the final outbound dial-peer is matched for the phone I am calling.

This issue only occurs with VoIP dial-peers, which is a big concern for me (users have not complained but I just don't like this delay) as the CME's have a VoIP dial-peer for the PSTN gateway, so the delay also affects external calls for users that need to route to the gateway.

Under telephony-service I have configured the T.302 timer to be 4 seconds. I have also replicated this for all the FXS and E1 ports that are in use under the voice-port (just to cover off on the analogue services I have).

I have configured a h323 class that has been applied to all VoIP dial-peers, as follows:

voice class h323 1
h225 timeout tcp establish 3
h225 timeout setup 3

I have also configured the following global config:

voice service voip
h323
no h225 timeout keepalive
call preserve

Some debug output from the called router ( dial-peer 4 is my "default" incoming dial-peer - configured with incoming called-number . )

May 17 14:40:25: //-1/7C8DCE548644/DPM/dpAssociateIncomingPeerCore:
Result=Success(0) after DP_MATCH_INCOMING_DNIS; Incoming Dial-peer=4

May 17 14:40:35: //-1/xxxxxxxxxxxx/DPM/dpMatchPeersMoreArg:
Result=SUCCESS(0)
List of Matched Outgoing Dial-peer(s):
1: Dial-peer Tag=20009

I understand that if a shorter number matches the begining of a longer destination-pattern, the shorter number call will experience a delay based on the T.302 timer, but I have tried to mitigate this as best as possible with specific dial-peers and patterns that do not overlap.

Any help/info/guidance is as aways greatly appreciated.

Thanks!

shodown

I'm at cisco live, I can take a look when I get back.

cisco_nerd

Lucky you, cheers bud.