Transcoding

danielfoxwhite · ‎02-22-2018

I'm having a hard time understanding when transcoding is done on a local router rather then on a CUBE or server. I'll lay out a what happened recently and see if I can obtain answers through the scenario...

We had an employee call a cell phone and was connected but during the call it droped. We found that it was due to BGP routes going down on our MPLS but, while looking at this I discovered what I think was an issue. The callers phone was associated with a media resource group list that didn't contain transcoding configuration. Normally we have a transcoding (xcode) profile in every MRGL that is associated with the local router. In this instance the callers Device Pool was assigned a MRGL that did not have an xcode configuration for the router that would be used. I found that not only was the xcode configuration not configured within Call Manager but the DSPprofile on the router being used was shutdown. So, here comes my questions...

1. How did the call ever become connected if no transcoding was configured? Where is the transcoding being done? I assume the cell carrier was not using G729 and some sort of transcoding had to be done.

The MRGL did have an MTP assigned but it was server based and we lost connection to the servers during the downtime. If no transcoding configuration exists does MTP do the translations between codecs or do we think that the transcoding was done on the carrier CUBE router?I doubt this makes total sense but hopefully someone can help me understand.

R0g22 · ‎02-22-2018

Transcoding always requires DSP's resources which are present on a IOS GW. Transcoder can be invoked by either CUCM (SCCP) or by a CUBE (LTI) when a codec mismatch is detected.

MTP's can only transcode b/w g711ulaw and g711alaw. Nothing else.

You would typically have transcoders/MTP local to the remote sites that require them because you don't want the RTP to go to your main HQ site and back over a WAN link causing BW issues. If you have tons of BW available then well you can get away with that as well.

danielfoxwhite · ‎02-22-2018

Transcoding always requires DSP's resources which are present on a IOS GW.

The router had DSP resources but the DSPfarm that specified transcoding and MTP was shut down. Would the call still utilize those resources with the profile shutdown? Here is what it looked like

dspfarm profile 10 transcode

codec g722-64

codec pass-through

codec g729r8

codec g711ulaw

codec g711alaw

codec g729ar8

codec g729abr8

maximum sessions 40

associate application SCCP

shutdown

You would typically have transcoders/MTP local to the remote sites that require them because you don't want the RTP to go to your main HQ site and back over a WAN link causing BW issues. If you have tons of BW available then well you can get away with that as well.

We currently only have transcoding profile set at our remote site. Does this mean part of the call will be traversing the WAN for MTP? How does that work?

R0g22 · ‎02-22-2018

Media resources such as an MTP/transcoder they need to register to a CCM via SCCP (as you can see from the "associate application sccp" command in the profile). If a resource is not registered, MRM process on UCM will not be able to use it/invoke it. Since this profile is shutdown, it won't be registered and hence can't be used.

If you have resources at remote site available locally, it is good. Transcoders are invoked by devices that do not support a requested codec/BW. When deploying your UC infra, you need to account for what codec you need for LAN and for WAN, which devices do not support which codecs etc. If there is a site which does not support a codec or let's say you have some third party devices which do not support a codec that are a mandate in your organization/deployment/design, it will require a transcoder to interwork the calls. You need to make them available locally and not over a WAN to let's say your main HQ site.

https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/collab10/collab10/models.html#20867

https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/srnd/8x/uc8x/media.html#wp1046264

danielfoxwhite · ‎02-22-2018

If a resource is not registered, MRM process on UCM will not be able to use it/invoke it. Since this profile is shutdown, it won't be registered and hence can't be used.

So in my situation, since the profile was shut down, how was transcoding being done?

R0g22 · ‎02-22-2018

It might have been a resource on another router or there might not have been a need for transcoding all together.
How does your region settings and your CUBE VoIP dial-peers codec configuration look like ?
Check the region relationship b/w your IP phone and the SIP trunk for this site that connects to the CUBE. That should give you more insight as to what you have configured and whether a transcoder would actually be required or not.