Crawling BitTorrent DHTs for Fun and Profit (2010) [pdf]

jhalderm.com

55 points·tikwidd··7 comments

Hi, I co-wrote this paper. Did not expect to see renewed interest!

FWIW, ClearView, the crawler used in the paper, almost certainly would not work unmodified today. The Vuze folks implemented Sybil attack defenses after our predecessor to this paper, "Defeating Vanish with low-cost Sybil attacks against large DHTs", was released. You can now only insert a finite number of nodes into the DHT per /24 you have (and a couple other flavors of subnet as well, probably /16 but I haven't checked the Vuze code in years). The last time I looked, IPv6 Sybils were still a possibility because it was unclear how much of the IPv6 space everyone would get.

Really cool idea.

Using this paper and the referenced one at http://arxiv.org/pdf/1009.3681.pdf%3Forigin%3Dpublication_de... I was able to make a proof-of-concept passive DHT crawler in a few hours.

link or it didnt happen

It's not in a releasable state yet.

You can do it yourself after reading the papers, though. I used node with the bittorrent-swarm, dht, ut_metadata and parse-torrent to build it.

Seems to be down.

Github mirror: https://github.com/ntoll/p4p2p/blob/master/papers/dht-woot10...

From the project: https://github.com/ntoll/p4p2p/

AFAIK, p4p2p is unrelated to the source code from this paper. A very quick skim of its github seems to confirm this.

Yeah, I meant the link was from that project - not that they are related.

Reply on news.ycombinator.com