Call Quality Issues

[gsitsupport]

We have are having multi customers who have issues making more than 3 calls. After 3 calls call quality goes into really bad jitter. We have even experienced it for ourselves. We have an 8 call 3cx system running as an Hyper-V VM 4 CPU 4GB RAM 100GB HD space on a dedicate 1GB nic, using a SOPHOS Firewall in a Co-location where we have dedicated 100mb ethernet.

We have 10 Extensions in our office and 4 remote extensions, however we have see the jitter start to happen when a call comes into the Ring Group.

Our office has 25mb up/down with FIOS and we use about 2mbps at max.

I am hoping (reaching) someone can help us understand if we need to change codec's or something to resolve this issue.

We are using Yealink T28p phones.

 

[markshehan]

Your specs look good.

Double check your actual bandwidth as something may be hogging it.

Do you have QOS turned on on the SOPHOS? Can you bypass the SOPHOS for the RTP (should be no need to inspect voip rtp packets)

You say that multiple customers have this problem with 3 calls but you only talk about 1 machine. Is this a typical example or are many users on the one vm?

 

[gsitsupport]

Hi

Thanks for the info.

Internally we use 3cx and have noticed that the jitter happens when 3 or more calls come in. We have seen this with other customers who have there own networks as well and they get jitter around the 3 call mark as well.

We are questioning whether are not there is a 3cx software setting we are commonly not implementing that would cause this to happen.

Also, would you recommend that setting up a VLAN would also increase performance here?

 

[markshehan]

What codecs are you using (in preference order) with your voip provider (and are you all using the same provider).

But it still doesnt make sense at 3 on all your customers. There is no setting in 3cx.

You can look at CPU load because 3cx delivers the audio for external calls.

If you have 3 internal calls at the same time and you still had jitter then that would imply a switch or similar as internal calls go direct to the extension and not via 3cx (with the standard setup).

For G711 you need 100K up and down (as a rule of thumb) so you would run out of licenses before bandwidth in your setup.

Is the jitter a problem both ways or just one way?

And you are not changing the default RTP ports in the settings are you? They should be 9000 to 9049 which allows for 25 calls then it will recycle

 

[i3]

I have a client who is experiencing a similar issue. I have tried everything to get this resolved. This particular client is using a Grandstream gateway with a full T1. The quality issue seems to happen for internal <--> external calls and not for internal <---> internal calls. I have provide 3cx support with call logs and pcaps with no success. They did make a suggestion to migrate the installation to a physical machine, which I did and seemed to have fixed the issue, however this is only a temporary solution as the client has invested money in a virtual Hyper-V environment and would like to keep the system on it.

I should add that our phone system is on a dedicated poe 10/100 switch with the phone system, no other traffic is on this network.

Please feel free to contact me, hopefully between the two of us we can find some similar environmental variables and figure this out.

Thank you,
Giuseppe

 

[markshehan]

I have a client who is experiencing a similar issue. I have tried everything to get this resolved. This particular client is using a Grandstream gateway with a full T1. The quality issue seems to happen for internal <--> external calls and not for internal <---> internal calls. I have provide 3cx support with call logs and pcaps with no success. They did make a suggestion to migrate the installation to a physical machine, which I did and seemed to have fixed the issue, however this is only a temporary solution as the client has invested money in a virtual Hyper-V environment and would like to keep the system on it.

I should add that our phone system is on a dedicated poe 10/100 switch with the phone system, no other traffic is on this network.

Please feel free to contact me, hopefully between the two of us we can find some similar environmental variables and figure this out.

Thank you,
Giuseppe

So 3CX support correctly identified the source of the problem - the VM you had it on. When you do internal to internal calls the audio goes directly between the two phones and nothing to do with 3cx. But when you make an external call the audio goes through the 3cx pbx. So now you introduce extra variables and workload that the 3cx server has to deal with and in your VM environment you have a problem.

So you have to look at the difference in the physical machines (the one it is on now and the bare metal hypervisor). The difference between the vm resources allocated and the current machine. Sound cards, drivers, memory, disc space, nic's. VM priority. There are hundreds of variables.

The best way for you to do it is to set up a VM that is as identical to the current machine as possible in terms of resources (Cpus, ram etc). Install the free version of 3cx on it and a sip trunk and put an extension on it. Then make external calls and change vm resources etc until you get it right.

Once it is correct you can then import your main 3cx and activate the license again and point the phones to it. And it should work