VCM being used even though Direct Media Path enabled

[john3voltas]

Hello,

== TLDR ==
Roughly half of the calls on this customer PBX are routing via VCM and the other half routing via direct media path for no apparent reason.

== Now some more details ==
One of these days, while looking at SSA, we detected that most calls were being routed via the VCM. Which seemed odd, because all trunks and extensions have "allow direct media path" enabled.
Giving it a closer look, we detected that even internal calls were being transcoded by the VCM/Jayde. Then we decided giving it an even closer look, and SSA does indeed report direct media path calls too.
So, it started looking a bit like 50/50, half using VCM and another half going DMP.

Running some quick tests with a couple of extensions we noticed that this happens not only with external calls (coming through SIP trunks) but also with internal calls too.
A couple of internal extensions, connected to the same LAN switch, on the same VLAN, thus on the same subnet, roughly half the calls between those 2 extensions will route via DMP and the other half through VCM.
Bear in mind this happens not only with these 2 extensions. This is a system-wide issue.
It is not the voice compression resources that we're worried about. It's the bandwidth. These 2 sites have 600 extensions on the IP500v2 side and they are all registering to the other site and RTP traffic is also routed via IPO server on the other site.

A few more details about the system.
Customer has 2 sites with a virtual SE R11.1.2 on one of the sites, an IP500v2 on the other site, a couple of SIP Trunks on each site and a mix of H.323/SIP Avaya endpoints, all registered against the SE server. Most endpoints are 1608's, 9620's and J139's and they are (unevenly) divided between both sites.
LAN switching is provided by Aruba 6000 series and Firewalling by Fortinet FG60, on both sites.

Any idea why this would be behaving like this?
Thanks in advance.
Cheers

 

[Westi]

So just to repeat this.
The same 2 extensions make calls back and forth and some are using VCM's and some not? And of course other extensions as well but you can recreate it with 2 extensions.

That is more than odd.


Joe
FHandw, ACSS, ACIS

https://www.millsidetc.com/

 

[john3voltas]

Yes, Westi.
This can be reproduced with whichever extensions I want.
I can make 10 calls from ext. 101 to ext. 102. Roughly, 5 calls will go through VCM and the other half will do DMP.
How could I get evidence on why this is happening? What specific System Monitor filters/trace options should I enable in order to debug this?
TIA
Cheers

 

[sizbut]

No idea, but "allow direct media path" could be better labelled "try to use direct media". It has no guarantee, in fact the guarantee to try and ensure successful calls is to always fallback to indirect media if there is any doubt.

We would really need more details about some of the individual calls where you expected direct media and you didn't get it. Then we would be looking for issues such as VoIP settings differences, routing issue, call features in use.

Also, are these calls remaining indirect for the whole call or just for parts of the calls. There are situations where the IP Office will temporarily turn a direct media call into an indirect media call because one of the user has done something that requires audio from the IP Office (tones, hold music, voicemail callflow prompts) or requires the IP Office to access the audio (conferencing, recording).





Stuck in a never ending cycle of file copying.

 

[Westi]

all IP to IP calls use a VCM channel to initiate the call but then should drop it once the audio gets routed directly.

Maybe the system rolls dice now to figure out how it works aka Avaya R11.1 new feature

Joe
FHandw, ACSS, ACIS

https://www.millsidetc.com/

 

[john3voltas]

@sizbut
Hi there.
Yes, I've always pictured "allow DMP" exactly that in that way, as in try to use DMP. I guess no one can force DMP. DMP simply is allowed or not. When not allowed, we are forcing all calls to use VCM resource even if DMP would be available.

But in this case it seems very weird. Because I can make 10 calls from ext. 101 to ext. 102 and roughly 5 calls will go through VCM and the other half will do DMP. Bear in mind that both 101 and 102 are connected to the same switch, same untagged VLAN on those ports.
The IPO server is in fact in a different VLAN, accessible only through WAN in a different country, it's the IPO server that's providing the VCM/Jayde resources.
The calls are just that. From ext. 101, dial 102, pick it up, talk, hangup. Still half do VCM and half do DMP and this is driving me insane lol.
Cheers