What is the default Jitter buffer size on ProSBC instances? And is it possible to adjust?
One of our customers is using a CaaS provider who has massive jitter (>1000ms) for a couple of seconds when the call is established.
When ProSBC receives RTP packets with a high percentage of sequence errors, we have found that that it stops forwarding this media, leading to one-way speech. Even if the sequence errors normalize after some time, the one-way speech condition persists.
The provider feels the problem is not on their end (they are wrong, because a packet capture from their own infrastructure has the sequence errors [screenshots below]). While we can probably ‘fight’ it over, we also want to assist our customer.
If there is a possibility to increase the Jitter buffer size, we could give that a go.
Looking at the data:
Max Jitter: 1130ms (The worst-case delay variation)
Mean (Average) Jitter: 134ms (The typical delay variation)
Delta: 20ms (The consistent time between packets being sent)
Packet Loss: 0%
It looks like there is a chronic network congestion along the path.
An average jitter of 134ms is about 6-7 times higher than what is acceptable. This can cause major audio problems. The massive spike here is the max jitter. It occurs when a network device (router, switch, firewall) is so overloaded that its queues fill up. Packets at the back of the queue are held for over a second before being forwarded, causing them to arrive drastically out of order and be flagged with “wrong sequence number.”
Does this issue happens on every call? You can verify the network path between the two endpoints.
The settings that you mentioned applies for a TMG unit but not the ProSBC. We have updated the page to avoid confusion. By right, it doesn’t look like adjusting the jitter buffer is the cure to this issue.
