Help with investigating/measuring lag in internal call

General Help
1

Hi there,
I am new to all of this and before embarking on a full IP set up, I am trying to experiment a bit with FreePBX and softphones.

I had everything working pretty well pretty quickly, but when making internal test calls we readlised that there was a really disturbing lag in the calls (delay betwee person A saying something and person B hearing it).

This is making conversations really difficult : pauses in the conversation, difficult to barge in (I am sure you knos what I mean).

This was confirmed with echo tests (*43) where the delay seems to be around 250 to 500 ms (wild guess, but this is sounding like ages)…

I have tried to record the echo test call, but it would not let me (I was able to record calls, so I know recording works).

My first question is : how can I measure this lag in order to help you telling me how bad/normal it is ? Recording the echo test would have been perfect has, making a “plop” I would see 2 spikes and would be able to measure time in between.

My second question is : how can I fix it and what can I expect with the real system (yes, ultimately, this is what maters) ?

My setup :

  • Asterisk 1.8.11.0
  • FreePBX… not sure but it is the last one.
  • Server is virtualized on Debian Squeeze
  • network with 15 desktops
  • ping to server is <1ms with the odd bump between 50 and 100.
  • used X-Lite as softphone.

I am really worried that it will not be better with real phones (leaning towards Yealink) and some users will be on softphone anyway.

The hard phones will have their own network and switch, but the server will still be in the data lan and the users on softphones will be on the network they are today…

Any help will be much apreciated.

JM

PS: sorry for the long post and my poor English.

2

The paid version of x-lite used to have call recording.

Not sure if the current Bria does or not

3

What is the delay if you ping the server from the network the phones are in?

4
  • ping to server is <1ms with the odd bump between 50 and 100.

Many thanks
JM

5

Well something is inserting the delay.

What does Asterisk show in ‘sip show peers’ command?

Also, is Asterisk running on a physical or virtual machine?

6
  • Server is virtualized on Debian Squeeze

Here is the sip show peers output :

sip show peers
Name/username Host Dyn Forcerport ACL Port Status
2010/2010 192.168.144.77 D A 17392 OK (15 ms)
2011/2011 192.168.144.55 D A 53835 OK (31 ms)
2020/2020 (Unspecified) D A 0 UNKNOWN
2021/2021 192.168.144.65 D A 43776 OK (3 ms)
2030 (Unspecified) D A 0 UNKNOWN
4010 (Unspecified) D A 0 UNKNOWN
4011/4011 (Unspecified) D A 0 UNKNOWN
4020/4020 (Unspecified) D A 0 UNKNOWN
4030 (Unspecified) D A 0 UNKNOWN
OVHtemp/0033972282097 91.121.129.17 N 5060 OK (104 ms)
10 sip peers [Monitored: 4 online, 6 offline Unmonitored: 0 online, 0 offline]

I am using ext 2010 (PC) and 2011 Android SIP Phone, the other “tester” is on 2021 (Android SIP phone).
OVHtemp is my provider’s trunk (no dedicated link yet).

I would understand some lag on external calls, but it really seems worse internally.

7

Your delay is virtualization guaranteed. Install on dedicated hardware and test.

8

Yes, Tony is right. It’s a virtualization issue.

9

Even though my ping is sub 1ms (most packets) ?

JM

10

Yes, real time software (such as Asterisk) and virtualization don’t work well together.

The folks that have this working have spent many hours tuning.

As far as Debian, now you have taken in outside anything I have any knowledge of.

11

Ok, will try on dedicated server next week.

Any idea how this can be measured ? Could try to record mic output and headphone output mixed together, but I have no idea how to do that…

Thanks again.
JM