Consider Really Stupid Routing (RSR), where each request or insert is forwarded to some random neighbor.
Assuming all y nodes cache data on insertion. We can caculate the chance of a single request returning data from a single insert: x = HTL insert y = HTL request N = number chance one not one particular node =(N-x)/N c = chance request is found = 1-( (N-x)/N)^y sample numbers: x=20 y=20 N=1000 c=33.2% N=10000 c=3.92% Ok, so why the hell am I posting this? No, I'm not proposing RSR instead of NGR. I'd just like to make a few observations: 1) c is a lot higher than you'd think. 2) c is a floor, if what ever routing we use is not much better (or worse) than c, we're doing something wrong (stupid). 3) "Right, but #2 requires you know N" If we did implement some kind of "route me stupid" message, we could do some statistics to get a pretty good estimate of N right? It seems this wouldn't violate annoymity any more than a regular insert/request. How do we count N now? I think I remember somebody counting the number of new version downloads. What's your favorite guess for N? I like 10000 cause it's round. If there is feedback between the number of people using freenet and the peformance, then it's hard to gauge wether things are better because you have less people using(or upgrading) freenet or it's actually improving. And Yes, I know if NGR fixes everything there's no reason to bother measuring peformance since it will be great anyway. __________________________________________________________________ Gesendet von Yahoo! Mail - http://mail.yahoo.de Logos und Klingelt�ne f�rs Handy bei http://sms.yahoo.de _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
