[p2p-hackers] Using UDT for swarming

[Greg Bildson]

I was looking around for these kinds of things last spring but didn't find
anything that great.  Given that this submission is dated August 2004, I
guess this proposal didn't exist at the time.  The implementation of our
reliable UDP with hole punching protocol was basically done by August 2004.

This proposal looks interesting.  I'm sure it is much more rigorous than
what we did but at the same time, the protocol overhead looks fairly
substantial.  What we did can also coexist with UDP delivered Gnutella
messages - for better or worse.

Thanks
-greg

> -----Original Message-----
> From: p2p-hackers-bounces at zgp.org [mailto:p2p-hackers-bounces at zgp.org]On
> Behalf Of Serguei Osokine
> Sent: Tuesday, January 25, 2005 1:33 PM
> To: p2p-hackers at zgp.org
> Subject: [p2p-hackers] Using UDT for swarming
>
>
>
> 	People, does anyone have an opinion about UDT?
>
> http://www.ietf.org/internet-drafts/draft-gg-udt-01.txt
> http://www.ncdm.uic.edu/papers/udt-protocol.pdf
> http://www.ncdm.uic.edu/papers/udt-control.pdf
>
> - at the first glance it looks like it might be pretty relevant for
> the swarming transfers, since it tries to maintain the fairness in
> the bandwidth allocation between multiple data streams with very
> different RTTs. That is something that TCP cannot do, and I have
> a suspicion that this issue might be one of the reasons of why
> BitTorrent works wonderfully when you have just one file to upload,
> but as soon as you get multiple simultaneous uploads, thing go sour
> pretty fast (i.e. Exeem and most of the P2P apps) - you start seeing
> very slow streams and such.
>
> (Actually, I'm mentioning swarming just because it tends to increase
> the number of concurrent uploads system-wide; even without it, the
> concurrent uploads have always been a problem in any P2P system I can
> think of, with a possible exception of Onion Networks - I'm not sure
> how well does Justin handle concurrency.)
>
> 	Anyway - did anyone try UDT or at least looked into it? Greg,
> LimeWire uses something much more simple, correct?
>
> 	Best wishes -
> 	S.Osokine.
> 	25 Jan 2005.
> _______________________________________________
> p2p-hackers mailing list
> p2p-hackers at zgp.org
> http://zgp.org/mailman/listinfo/p2p-hackers
> _______________________________________________
> Here is a web page listing P2P Conferences:
> http://www.neurogrid.net/twiki/bin/view/Main/PeerToPeerConferences