On Wednesday 02 July 2003 07:38 am, andrew luecke wrote: > I'm just wondering if any research has been done into whether or not > freenet is safe against algorithmic complexity attacks. > > An algorithmic complexity attack is where u consume a computers resources > by generating a worst case hash table, which in a hash table which uses > linear probing is eventually the entire hash table, and one which uses > chaining, the max length of chain. This means that by flooding the network > with appropriate insertions to generate this worse case scenario, every > time for instance u searched the last key on the chain, of a 50gig > datastore, u would have to search thru 50gigs of data.... which would be > catestrophic for the node, and they would probably have to wipe their data > store.. while I havent seen Freenets implementation of a hash table yet, I > am wondering if there is defense against such an attack, and if the hash > table can survive it. There is a massive difference between avg case and > worst case time in most hash tables, and thats why its a risk to a server. > > Just wondering if theres defence against this.. Hoping that defence is > built in too to prevent a sites.ref from having multiple entries of a > computer , with multiple ports connected, to prevent possible sparseness > attacks (to make freenet so sparse that its too broken to be used)
Correct me if I'm wrong, but doesn't this attack involve the generation of the hash key. If that is the case then Freenet is completely invulnerable because the requesting or inserting node provides the key that is to be used for the data, and if there is a collision it is assumed to be thee same data. _______________________________________________ devl mailing list [EMAIL PROTECTED] http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl
