[cisco-voip] MTP resources Codecs mismatch and DTMF relay

[Heim, Dennis]

You mention enabled MTP on those SIP trunks, you can enable it at the trunk or gateway and it will force every call to use MTP. However, if you set it, I believe in the trunk profile, or security profile, cannot recall exactly, but you can set it to only engage the mtp resource if necessary.

Dennis Heim
Sr. UC Engineer
World Wide Technology
Office: 314.212.1814
Email: dennis.heim at wwt.com<mailto:dennis.heim at wwt.com>
www.wwt.com<http://www.wwt.com/>

From: cisco-voip-bounces at puck.nether.net [mailto:cisco-voip-bounces at puck.nether.net] On Behalf Of abbas Wali
Sent: Tuesday, September 25, 2012 1:55 PM
To: cisco-voip at puck.nether.net
Subject: [cisco-voip] MTP resources Codecs mismatch and DTMF relay

Recently we had a major VOIP fault which was escalated to TAC. Whom also spent some good time before realizing that we were running out of MTP resources at busy times.

What was happening that calls coming on phones and as soon as the receiver was lifted, the called party was hearing a fast busy.
Initially we had to restart the whole cluster and it used to settled down.

In the end we have to increase the MTP resources in the Service Parameters from 24 to 48 for each of the nodes, since then this fault has gone away.

Just going back into recent past -we believed that this issue started after we enabled MTP resource requirements for the SIP trunks, who at that time were struggling with the DTMF relays. So by enabling it for the SIP trunks we got into contentions for the rest of the network.

My question is (and I have overheard somewhere) that if your cluster is configured properly, you will never require the use of MTP resources (or may be in a very lesser amount). Also I feel like some mis configuration on the SIP trunks which has forced us to get help from the MTP resources  to support the DTMF relays.

Below are some of the points which emphasizes my points


 1.   PCM type for our E1 cards are set to A-Law but when we make an VOIP to VOIP call internally it shows the codec as ULaw. (so to me it looks like that we are using MTP to convert from A to U for incoming PSTN calls)
 2.   For the DTMF -the gateways are configured with  mgcp dtmf-relay voip codec all mode out-of-band while some of the SIP trunks are set to RFC 2833. I believe these are two different modes which again requires MTP for translation.

I believe if we make the codecs even by forcing the CUCM to use A Law internally as well and also use one single mode for the DTMF relays then we should not require to increase the MTP resources.


Just to add we are using CUCM 8.5. we have some 7k phones. we have MGCP gateways. And SIP trunks to 3rd party IVR's and another neighbour CUCM 7.
Any help will be appreciated.



--
@bbas..
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20120925/358e882d/attachment.html>