We're experiencing occasional call quality issues on our new 3cx system. I've followed the guidelines in the hyper-v instructions and also the server-sizing recommendations. The VM is running on dell 740xd servers with gobs of ram, CPU cores and multi-tier storage (including NVME and SSD). CPU utilization on hosts is under 20%.It's a very fast system. I've given 3cx 10GB of ram and 8 cpu cores etc. CPU utilization in the VM was never high when calls were bad. Regardless I've since set the VM to high priority and reserved CPU.
When we tried to port some numbers over with our telco the other day, the call quality was horrible and we had to roll back. Note that our internal calls are all perfect, including calls over MPLS WAN, that go from one coast to the other (North america). The only calls we were having trouble with were the inbound/outbound over our sip trunks where 3cx would be directly in the call path. We have the telco now monitoring the trunks but have found nothing so far.. They come straight into our building over dedicated fiber optic. From the PBX to the telco switch, it's 1GigE minimum (core is 10G). I have nics in the hyper-v servers dedicated for 3cx with no other traffic on them connecting straight into core, and core connects over a l2 trunk to a L2 switch that's connected directly to telco fiber switch. Very little room for things to go wrong. Traffic is all marked DSCP EF and I've confirmed packets going in and out with that marking. Again, there's nothing but voip traffic over this link and we're nowhere near saturating 1GigE. And there's no SBC between the 3cx system and the telco. There are no interface errors on any of the switch interfaces, no congestion, and no packet loss. So i'm at a bit of a loss at this point.
Last time i did any serious voip stuff was a decade ago and there were timing issues for asterisk unless you had dedicated hardware like a T1 or PRI for timing. I did not get the impression that there were any of these sorts of issues with 3cx, but I figured I would ask if there is anything else I could look at on the hyper-v side. It could of course be the trunk itself but we've had a second trunk over the same fiber hooked up to an SBC that bridges to our old PBX and have literally never had an issue there.. From a network perspective the connection seems clean.
Thoughts? Thanks..
When we tried to port some numbers over with our telco the other day, the call quality was horrible and we had to roll back. Note that our internal calls are all perfect, including calls over MPLS WAN, that go from one coast to the other (North america). The only calls we were having trouble with were the inbound/outbound over our sip trunks where 3cx would be directly in the call path. We have the telco now monitoring the trunks but have found nothing so far.. They come straight into our building over dedicated fiber optic. From the PBX to the telco switch, it's 1GigE minimum (core is 10G). I have nics in the hyper-v servers dedicated for 3cx with no other traffic on them connecting straight into core, and core connects over a l2 trunk to a L2 switch that's connected directly to telco fiber switch. Very little room for things to go wrong. Traffic is all marked DSCP EF and I've confirmed packets going in and out with that marking. Again, there's nothing but voip traffic over this link and we're nowhere near saturating 1GigE. And there's no SBC between the 3cx system and the telco. There are no interface errors on any of the switch interfaces, no congestion, and no packet loss. So i'm at a bit of a loss at this point.
Last time i did any serious voip stuff was a decade ago and there were timing issues for asterisk unless you had dedicated hardware like a T1 or PRI for timing. I did not get the impression that there were any of these sorts of issues with 3cx, but I figured I would ask if there is anything else I could look at on the hyper-v side. It could of course be the trunk itself but we've had a second trunk over the same fiber hooked up to an SBC that bridges to our old PBX and have literally never had an issue there.. From a network perspective the connection seems clean.
Thoughts? Thanks..
