vicidial.org

We are facing an issue with our goautodial/vicidial instance on owned hardware. Even on low load and icmp roundtrips of like 20ms, when a user calls inbound onto our system from their tradidtional ISDN system, then the agent transfers back to an extension of the customer with that ISDN system, a latency of around 600ms is noticed. When they put their fairly new Siemens ISDN/VoIP system in place, this same route creates even a 2000ms latency, making it useless. We are facing these useless calls with dozens of inbound clients, which are all using some kind of newer ISDN to VoIP system. But, the standard latency of 600ms that we get when our system and only traditional ISDN is involved is even way too much.

We've been digging around, trying, failed succeeded in some ways and now we're down to Meetme as being the trouble causer. Can anyone confirm this would probably be the case?

And if you're almost sure that we have to replace Meetme, would (or could) configuring asterisk 1.6 with our instance, changing it to use Conference instead of Meetme be the solution to the timing issues that we face?

After almost a year and several clients' clients compaining big time, it is time to solve te problems indefinitely.

We tried app_conference a few years ago and had issues with audio quality at high load compared to using meetme with a hardware timing source.

We are working on Asterisk 1.8 support, but we will not be supporting Asterisk 1.6 because it is already end-of-life.

Ok, thanks! So, in your opinion it is because of the Meetme mixing, app_conference won't fix it (saw vici integration on voip-info, I think you might have written that), Asterisk 1.6 is not an option and Asterisk 1.8 is not coming yet.

That'll mean we have just got to choose between "won't even try" and "don't even start"? Isn't there any way to fixing this?

I have never heard of MeetMe causing 580ms of delay. The fact that changing exterior equipment even effects the latency points to something outside the asterisk box.

What are the specs of your server?
What are you using for a dahdi timing source?
What are you using for network equipment?
Is there QoS on the network? What happens when you turn it off?

I have seen switches and the like that prioritize ICMP ping over other traffic so ping times look fantastic, but the latency for other services is junk.