Question:
If I am IAX trunking between 2 Asterisk instances, and ultimately connecting
to SIP endpoints on BOTH ends of the call, can I let the ENDPOINTS do ALL
the jitterbuffering, or must the iax-trunk do its own jitterbuffering?
I'm asking because I'm ignorant to the nuanced MECHANICS of the
transport:
That is to say, if asterisk is passively sending voice frames from one
protocol the other, then it clearly WOULD NOT matter if they go through the
asterisk instance out-of-order. The endpoint's local jitterbuffer can
re-order the frames/packets. Therefore in that scenario, it would seem that
one could effectively eliminate the IAX jitterbuffer entirely and slightly
decrease latency.
On the OTHER hand if the voice frames are being 'repackaged' by asterisk
on
new time bounaries, then naturally iax would need to do ALL of its own
jitterbuffering to prevent incremental losses from out-of-order packets.
As I write this, it occurs to me that there may be a third option in which
IT DOESN'T MATTER because there will be little or no out-of-order delivery
within the local ethernet broadcast domain (to which each sip endpoint is
connected), AND THEREFORE the IAX de-jittering would effectively cause the
AUTOMATIC jitterbuffer on the endpoints to 'dry up' appropriately to
near
zero.
Could someon critique my logic or speak to this question?
Thanks!
-Karl