Need technical help on an RTP political issue

Hi all,

I’m hoping for some technical arguments to help on an issue which is rather political within the large organization in which I work.

We have an SIP trunk connection from a major carrier ingressing our network at an SBC, and ultimately the work is done at a SIP server managing the endpoints.

We set up G.711u calls with a packetization period of 20ms.

From our network’s perspective, inbound RTP is received at regular 20ms intervals. However, outbound RTP is being put on the network by the SIP server at irregular intervals. Some intervals are 30ms, and some are near-0ms. It’s a pretty regular pattern: 30ms, 0ms, 30ms, 30ms, 0ms, repeat.

This obviously is not normal behavior. The carrier is refusing to certify connections until this is resolved. And our SIP Server vendor is insisting that it isn’t breaking any standard. I can’t really find any clear arguments.

Any help would be appreciated.



Appears to be off topic.

That pattern isn’t sustainable, as the overage interval is less then 20ms, so it is guaranteed to overrun any latency buffer. Are you sure the objection is to the jitter, rather than the average rate?

If it did balance, it would be a tolerable jitter for for a poor network connection, but unreasonable for a timing source. RTP concentrates more on how to cope with jitter than specifying limits for it, but I would say that the network operator had reasonable, if not statutory, grounds for saying it was broken.

If this were Asterisk, and the average rate was OK, I would assume someone was trying to run it in a virtual machine or on a strange OS, in both cases, with scheduling intervals of 30ms.