5 Dec
2014
5 Dec
'14
12:14 p.m.
On 05/12/14 04:38 AM, Juha Heinanen wrote:
what has been the motivation that rtpengine control
protocol is using
udp instead of tcp or ssl?
we have noticed that reply to delete containing stats is too large for a
single udp packet over ethernet and causes udp fragmentation and
fragmentation needed requests.
Because that's what the original rtpproxy module used, and it seemed to
work.
If it were to use TCP and if the stats reply is too large for a single
packet, the reply would still be sent in multiple packets. Not
fragmented, but TCP segments. Nothing gained unless you have packet loss.
cheers