Re: [BEHAVE] BitTorrent with shared IP addresses

Ok, I'm going to writte an I-D on this BitTorrent behavior.

Pierre 

-----Message d'origine-----
De : Dan Wing [mailto:dwing@cisco.com] 
Envoyé : vendredi 31 octobre 2008 17:56
À : LEVIS Pierre RD-CORE-CAE
Cc : behave@ietf.org
Objet : RE: BitTorrent with shared IP addresses

> -----Original Message-----
> From: pierre.levis@orange-ftgroup.com 
> [mailto:pierre.levis@orange-ftgroup.com]
> Sent: Friday, October 31, 2008 6:12 AM
> To: dwing@cisco.com
> Cc: behave@ietf.org
> Subject: RE: BitTorrent with shared IP addresses
> 
> The second can download the file when the first has finished and when 
> it has re-launched the bittorent download.
> After further investigation I saw in the BitTorrent seed (BitTorrent 
> 6.0.3)in Options/Preferences/Advanced, the parameter:
> bt.allow_same_ip
> When this parameter is set to false the behavior is as mentioned.
> When this parameter is set to true the two PCs sharing the same IP 
> address can simultaneously download the file.

Thanks for your investigation on this.  Would you be willing to write up an I-D describing this application behavior?  I expect we will find other applications that expect each 'user' has one IP address - which will no longer be true with IPv4 address sharing.  This affects NAT44 (CGN), port borrowing, DS-Lite, and IPv6->IPv4 translation.

I touched on this problem a little bit in http://tools.ietf.org/html/draft-wing-nat-pt-replacement-comparison-02#section
-9.1 but without specifics on BitTorrent, of course.

-d

> Pierre
> 
> -----Message d'origine-----
> De : Dan Wing [mailto:dwing@cisco.com] Envoyé : mercredi 29 octobre 
> 2008 19:22 À : LEVIS Pierre RD-CORE-CAE Cc : behave@ietf.org Objet : 
> RE: BitTorrent with shared IP addresses
> 
>  
> 
> > -----Original Message-----
> > From: pierre.levis@orange-ftgroup.com 
> > [mailto:pierre.levis@orange-ftgroup.com]
> > Sent: Wednesday, October 29, 2008 7:52 AM
> > To: dwing@cisco.com
> > Cc: behave@ietf.org
> > Subject: RE: BitTorrent with shared IP addresses [was RE: 
> > [BEHAVE] TR: I-D Action:draft-boucadair-port-range-00.txt]
> > 
> > Dan,
> > 
> > Yes, the test you carried out work also in our lab (I have
> tested the
> > same torrent download as yours on two different PCs sharing
> the same
> > IP address).
> > 
> > In fact, the result we have mentionned in our draft is from another 
> > test which is much more constrained:
> > 
> > "For one P2P (Peer-to-Peer) application tested, we found that two
> >    peers each behind a CPE sharing the same public address cannot
> >    download a SAME file from a source peer."
> > 
> > By "SAME file from a source peer" we mean a single file on a single 
> > location.
> > Then, the two PCs with the same IP address tried to
> download the file
> > from the very same location, by BitTorrent.
> >
> > So, of course we cannot say that it does not work at all,
> we can only
> > say that it does not work in these very particular conditions.
> 
> Ah, okay.  So the failure occurs when a file that has only one seed in 
> the BitTorrent swarm, like this:
> 
> 
>              PC-A--+
>                     \
>                      NAT--seed
>                     /
>              PC-B--+
> 
> 
> And PC-A is able to download the file, but PC-B can't download the 
> file.
> 
> What happens after PC-A completes downloading the file -- can PC-B 
> start downloading it immediately after PC-A finishes?
> Perhaps after the next tracker update, can PC-B start downloading 
> it?   Or does your testing show that PC-B is *never* able to 
> download the file because PC-A already downloaded it from the seed?
> 
> -d
> 
> 

_______________________________________________
Behave mailing list
Behave@ietf.org
https://www.ietf.org/mailman/listinfo/behave