Selection of IP media resources for SIP trunk

[rejackson]

Calls through our SIP trunks between CM and SM are using media resources at random remote locations. This problem is the same as thread690-1690521 but I have a lot more locations and network regions and it includes the voicemail system.

For example I am in Dallas in NR 10, the trunks to SM are in NR 2. I press my voicemail button and the trace shows my station in NR 10 connecting to the voicemail system in Dallas through NR 34 in Boston. It was consistant for several tests and when I removed the codec for NR 34 from NR 2 it changed to another remote NR. What I could not test at the time was whether the Boston office could reach voicemail when there was no codec for NR 34 in the NR2 configuration.

I need to make the SIP connections use the medres in NR 1. Is this possible? is it as simple as removing the codecs for the other NRs from NR 2?

 

[latitude12]

There are a lot of things to consider. To start with:
What type of gateways do you have?
What are the near end and far end regions for your signalling group(s) to SM?
Are you using IP or TDM phones?
If an IP phone, is it definitely registered in NR 10 (check 'list reg', is the phone IP in the network map)?
Do you have DSPs in NR 10?
Do you use a virtual region between your different sites, or are they all directly connected?

 

[kyle555]

Are you using intervening network regions? Do all regions have DSPs? What's probably happening is you have every region directly connected to NR1. If your SM trunks are in a NR without DSPs, then calls out that trunk will try to find DSPs from your NR to the the NR of the SM trunk. If all NR's are "directly connected" to all other NR's, then this is where you get your problem of unpredictable DSP selection. Leveraging intervening regions in your connectivity should make that a little more predictable for you.

 

[rejackson]

I dont think we have interveniing NRs. The remote locations each have an MG700 or MG430 in its own NR that connects to the central location. The signaling for the far end of the SIP trunks is in NR 250 not the NR 2 I was thinking. NR250 does not have any media resources assigned. I am not sure how the NR of the Near end is determined. It is a CLAN and the IP address is not in the ip network map.

 

[latitude12]

The CLAN NR is defined on the IP interface screen. As kyle555 has suggested, it looks like a DSP for another region is selected because all your regions are interconnected. Why exactly this happening isn't clear to me, but will be something like not having DSPs in a particular region, a codec mismatch somewhere that means the call needs to be transcoded, or you have direct media disabled somewhere (maybe on the SIP trunk or between some of your NRs). Unfortunately there isn't an easy fix, because even small NR Config changes can have unexpected consequences if you don't totally understand your configuration. I would suggest reading the NR Config Guide, particular the stuff about virtual (or intervening) regions. It's old (circa CM 3.0) but I've never seen a newer version and the basics haven't changed.

https://downloads.avaya.com/css/P8/documents/100011997

If you have MPLS (or any kind of shared WAN service) then virtual regions are essential to get proper CAC and ensure segregation of resources.