[OpenSIPS-Users] detecting mediaproxy across multiple opensips instances

[Iñaki Baz Castillo]

2009/6/9 Dan Pascu <dan at ag-projects.com>:
>> In case other proxy also tries to use RtpProxy for a request already
>> forzed by other proxy, then "force_rtpproxy()" will return "false"
>> since it detects the line "a=nortpproxy:yes" in the SDP so the second
>> Rtpproxy is not used and $rc is false.
>>
>
> Having to rely (and depend) on what other proxies do, for your own NAT
> traversal is a bad design decision.

If we consider the hypothetical case in which a a VoIP platform is
designed not just for PSTN or local-users-to-local-users calls, but
also for real outbound calls to other domains managed by other
proxies, then we cannot have control of how the NAT is handled in
those remote proxies. The only we can do is "fix" our side, which
involves "fixing" signaling and media when our user is behind NAT
(with no STUN and so).
This is: I would *never* rely on external proxies to handle my *own*
NAT issues (my users behind NAT), I fix NAT in my platform so when the
request leaves it, it seems to arrive from public IP (signaling and
media).

Of course, considering SIP just for a closed environments
(provider2clients and clients2provider) is a limited vision (even if
nowadays it's the real usage of SIP when involving end-users). This
scenario simplifies a lot the use case so handling two media proxies
for the same call is just *prohibited* by design :)

Regards.


-- 
Iñaki Baz Castillo
<ibc at aliax.net>