Serguei Osokine wrote: > On Monday, December 01, 2008 David Barrett wrote: >> Saw uTorrent switched to UDP: >> >> http://www.theregister.co.uk/2008/12/01/richard_bennett_utorrent_udp/ > > Is it just me, or this article is really unnecessarily alarmist? > Richard Bennett describes the situation as if there is no congestion > control in the UDP file transfer protocol used by uTorrent, which > I find a bit hard to believe. There is congestion control. And the idea with using their custom transport is that its congestion controller yields to TCP and uncontrolled UDP (VoIP etc). Ease of NAT traversal is more or less a freebie, although a nice one. Unfortunately I have a disclosure agreement with BT so can't share more than what you could find out with a managed switch, Ethereal and a bit of skilled sleuthing. But since I'm no longer in their employ, I can at least save you the effort. It's very much congestion controlled. > > Is this really the case? I cannot imagine how the data transfer > protocol without any congestion control can possibly exist Of course not. That would be asinine.
I started the code to do r-v NAT traversal in uT. Don't know if they use this or started over with a different scheme. NAT traversal is pretty easy, unlike UDP congestion control over the public Internet, which falls in the category of rocket science IMO. STUN isn't used, but could be employed to optimize r-v selection and skipping connections that can't succeed. _______________________________________________ p2p-hackers mailing list [email protected] http://lists.zooko.com/mailman/listinfo/p2p-hackers
