[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: distribution of per-port traffic across exit nodes
. . . and, just to throw another wrench into the mix, what happens if
port B doesn't actually get any traffic? Do routers A and C split it
evenly, or do they assume port B will get equal traffic, thus sticking A
with 2/3 of the traffic and C with 1/3?
And, more to the point, is there any way to collect bulk port usage data
from nodes without destroying anonymity? Because that seems to be the
ideal solution to this, but not without unfortunate side effects.
(And another question on top of this - should all traffic be considered
Geoffrey Goodell wrote:
Suppose that there are three routers (A, B, C), all with equal capacity.
Router A is configured to exit to port A, router B is configured to exit
to port B, and router C is configured to exit to both ports. Suppose
also that traffic through the Tor network is evenly distributed between
ports A and B.
How will Tor distribute exit traffic across these servers? In
particular, will router C receive more exit traffic overall simply
because it has a more liberal exit policy (e.g. routers A and B each
receive 1/4 of the total traffic), or will routers A and B take up more
of the slack since they have more conservative policies (e.g. routers A
and B each receive 1/3 of the total traffic)?