On 05/14/2012 10:44 AM, David Barrett wrote:
Oh interesting, so it's attacking the DHT layer in addition to the
Torrent layer? Clever!
In the torrent layer, any individual torrent client is able to
gradually weed out bad actors by identifying slow transfer speeds
and/or high rates of hash mismatches. But I'm not familiar with the
details of the public torrent DHT -- do individual nodes have any
protections in place to identify and ignore bad nodes?
Bittorrent DHT is based on Kademlia, and I am not an expert on
Kademlia (I will let my colleagues whose paper [2] I mentioned
earlier speak out as experts).
But, it seems to me that once the sybils have made themselves
responsible for the resource, they can selectively include good peers
but in smaller numbers so as to keep the rate of successful downloads
low. So the point is not to completely stop downloading but to impact
it enough that a majority of downloaders fail. A snipped from a
follow-up interview of Pirate Pay's founders appear to support this [1]:
"We used a number of servers to make a connection to each and
every p2p client that distributed this film. Then Pirate Pay
sent specific traffic to confuse these clients about the real
I.P. addresses of other clients and to make them disconnect
from each other,” Andrei Klimenko said. “Not all the goals were
reached. But nearly 50,000 users did not complete their downloads."
Though I imagine the cost to wage this attack is greater than the
benefit obtained, so the most likely (and most effective) response is
probably "do nothing".
The analysis by Varvello and Steiner [2] indicated that 3 or 8 sybils
are enough to command an info hash. I suspect that the cost, then,
from Pirate Pay's point of view is low and the perceived payoff high.
[1]
http://rbth.ru/articles/2012/05/10/russian_innovators_pursue_prototype_to_prevent_piracy_15605.html
[2] http://www.moritzsteiner.de/papers/dht_traffic_localization.pdf
Thanks,
- vijay
--
Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent
1960 Lucent Lane, Rm. 9C-533, Naperville, Illinois 60563 (USA)
Email: vkg@{bell-labs.com,acm.org} / vijay.gurb...@alcatel-lucent.com
Web: http://ect.bell-labs.com/who/vkg/
_______________________________________________
p2p-hackers mailing list
p2p-hackers@lists.zooko.com
http://lists.zooko.com/mailman/listinfo/p2p-hackers
