El Lunes, 3 de Noviembre de 2008, Andreas Sikkema escribió:
Compared to H.323 SIP is cobbled together by a couple of monkeys who only knew how to speak HTTP and had heard of a planet in a galaxy far, far away called Telefonis where people talked to each other using tin cans.
<ironic> Of course, that should be the reason why all the new implementations use H.323 instead of SIP. </ironic>
Calling protocol names is not productive. Separating media from the signaling might theoretically by a "correct" concept, in practice there's much to be said for combining them.
Well, if you think that combining signalling and media is a cool feature then I have no more to add. Yes, NAT exists and it's a pain, but that is not reason to go back to earliest years of the telephony. Maybe you think that IAX is better than SIP because "it's better for NAT", but that's not true: IAX requires signalling and media going together while with SIP you can decide it (you can use several methods to resolve NAT, like STUN, Comedia mode in UAS, RTP proxies...). If you have your proxy/PBX in Tokio and two subscribers talk in Australia: 1) With SIP audio can go directly between them (so no latency). 2) With IAX audio must go to Tokio and come back. Since we are speaking about realtime communications I hope you agree with me that second option is not very suitable, do you?
I agree that, anyway, SIP is complex, and the worst, it could be designed simpler. But at least, SIP is extensible and customizable, while IAX/H.323 is like a stone.
Regards.