We also need to support API for configuring stun server into the peer connection.
Cheers Suhas/crypt Sent from my iPad On Oct 31, 2012, at 3:35 PM, "Ralph Giles" <[email protected]> wrote: > (discussed briefly on irc, following up here for wider reference) > > The NAT traversal used by WebRTC requires in part a STUN (or TURN) > server clients can ask for their public IP address. So far, we've just > hard-coded a particular public stun server[*] into our code. > > While webrtc allows web content to specify its own stun or turn server, > I always expected we would need to deploy and maintain our own stun > servers as a default option. I've volunteered to get something set up > for testing. > > Because a round trip to this server is part of call setup, it's better > to have geographically distributed servers to reduce latency. I propose > we use Amazon's EC2 virtual machines for this. They're easy to get > started, Mozilla already has a relationship with them for other other > work, and they have data centres in six locations around the globe. > > I've volunteered to set something up for testing. STUN is a very simple > protocol, so this should be inexpensive and easy to set up. I've opened > bug 807494[+] to track this. Please contribute any technical ideas there. > > I wanted to mention the project here just so people are aware of it, and > to get feedback on the general idea. > > -r > > [*] e.g. > https://github.com/mozilla/mozilla-central/blob/fce8a2c197d97d0a8a443c4d59055d73b638c25f/media/mtransport/nricectx.cpp#L278 > [+] https://bugzilla.mozilla.org/show_bug.cgi?id=807494 > _______________________________________________ > dev-media mailing list > [email protected] > https://lists.mozilla.org/listinfo/dev-media _______________________________________________ dev-media mailing list [email protected] https://lists.mozilla.org/listinfo/dev-media
