On 04/08/14 21:15, Karsten Loesing wrote: > On 31/07/14 23:44, Nusenu wrote: >> Hi, >> >> I would have a question regarding the bw-flags graphs on metrics.tpo [1] >> >> "guard bw history" >> a) Does this include the entire accumulated traffic of relays having >> the guard flag? (which would include the traffic of a guard relay >> acting as a middle or exit relay - if it also has the exit flag) >> >> or >> >> b) Does this only and exclusively include the entire accumulated >> traffic that *enters* the tor network via all guard relays? >> (to actually do that the traffic accounting on relays would have to do >> accounting accordingly on a per connection type level) > > a) is correct. > > Minor nitpick: a relay can exit even if it doesn't have the Exit flag; > that flag is only assigned for load-balancing purposes. > > There's no data available for b). > >> Which would bring me to the next question (which would be required to >> actually do the required accounting for (b)): Is a relay able to tell >> whether it is being used as first or second hop solely by looking at >> packets (not their source)? > > Looking at the packet source and comparing it to the list of known > relays would probably be the most reliable way for counting bytes as > first or second hop. But this is not implemented, and there are plans > to implement this, to the best of my knowledge.
Erm, "there are *no* plans to implement this" is what I meant to say there. >> Same question applies to: "exit bw history" > > a). > >> Does your answer also apply to [2]? > > Yes, except that [2] distinguishes between Guard only and Guard & Exit. > > All the best, > Karsten > >> [1] https://metrics.torproject.org/bandwidth.html#bandwidth-flags >> >> [2] https://metrics.torproject.org/bandwidth.html#bwhist-flags > -- tor-talk mailing list - tor-talk@lists.torproject.org To unsubscribe or change other settings go to https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk
