[Standards] Do we need STUN?

Robert McQueen robert.mcqueen at collabora.co.uk
Fri Mar 9 00:33:11 UTC 2007


Greg Hudson wrote:
> If it is ever going to be possible to bridge SIP and Jingle, SIP and
> Jingle endpoints need to be using the same NAT traversal algorithms,
> right?

Right, unless you relay all the media too.

> So even if we can design something simpler than STUN/TURN by taking
> advantage of the XMPP server, we'd be unnecessarily consigning ourselves
> to a non-interoperable silo in the VOIP world.

Yes, and direct peer-to-peer interoperability is something I'd like to
aim for if at all possible.

> (This is assuming we haven't already consigned ourselves to a silo.  My
> understanding of both SIP and Jingle is only ankle-deep at the moment.)

Well we havn't yet. libjingle doesn't play ball with anyone else, but
there's nobody else to play ball with yet. A Jingle implementation that
uses "raw UDP" will be trivially gatewayable onto a SIP client, and I'd
like the same to be possible for ICE-enabled clients too.

Regards,
Rob



More information about the Standards mailing list