Alex Pankratov wrote:
>
> It appears to be a good time to mention BitThief, which allegedly
> defeats BT's tit-for-tat mechanism.
>
> http://dcg.ethz.ch/projects/bitthief
Thanks for the reference! I just read through it, and it is merely using the
"optimistic unchoke" slot. (In Mojo Nation days, we called called this attack
"flaming smurfs". I have no idea why -- it was Jim McCoy's terminology.
I guess individual Smurfs are small-time, but lots of them can overwhelm you?)
(It is also taking advantage of the altruistic contributions of seeders, of
course.)
Bram was well aware of this issue from the start, and his design has always
defended against it. To see how robust BitTorrent is against this attack,
observe that no matter how many BitThiefs there are out there, they will be
competing against one another for the fixed number of optimistic slots. As
long as there is only one BitThief then it is only a bit slower than a normal
client (one to four times as slow, approximately, depending, see the paper for
details), but the more BitThieves there are the slower each one goes.
The analysis in the BitThief paper ("Free Riding in BitTorrent is Cheap" by
Locher et al., URL above) is limited to the case of one BitThief in a network
full of standard clients (either mainline or Azureus), and is thus not useful
for predicting the effect of BitThief on the real life BitTorrent networks.
Regards,
Zooko
_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers
