|
Title: High outdegree network attack Post by: misterbigg on February 19, 2013, 04:08:48 PM From my understand of how the overlay network works (which is identical to that used in pre-Ultrapeer Gnutella implementations), the more peer connections maintained, the higher the amount of duplicated transaction and block messages due to the flooding algorithm. Is it possible for a malicious miner with a lot of bandwidth to degrade the network by maintaining a huge number of connections to peers? The goal would be to slow down other miners so they produce orphans more often.
Right now blockchain.info maintains around 3,000 connections does this increase the amount of wasted bandwidth due to duplicate peer messages? Title: Re: High outdegree network attack Post by: misterbigg on February 19, 2013, 10:38:09 PM Is this just an imaginary problem?
Title: Re: High outdegree network attack Post by: mc_lovin on February 19, 2013, 10:45:46 PM So we need a p2p blockchain.info..
Title: Re: High outdegree network attack Post by: gmaxwell on February 19, 2013, 11:49:32 PM Right now blockchain.info maintains around 3,000 connections does this increase the amount of wasted bandwidth due to duplicate peer messages? I noticed them taking up a bunch of slots on my own nodes (e.g. 6 total sockets, across multiple) and blocked them. Nice casual privacy increase too. ... and their behavior strikes me as a bit unethical too— wastes your resources, in order to compromise your privacy, which you can get back using a service they sell you. Lame.So, yes, sure. It wastes resources but it's not a major concern unless its replicated by more parties. The motivation you describe doesn't follow— any sane large miner does their mining from a shielded node with low degree— connecting only to their own nodes, other known miners and major nodes, and they separately run large public nodes. |