http://torstatus.blutmagie.de indicates that only 21.4% of Tor nodes are exit nodes. Are we wasting this precious resource by running non-exit traffic through these nodes?
-Pascal
On Sat, Aug 31, 2013 at 11:32:55AM -0500, Pascal wrote:
http://torstatus.blutmagie.de indicates that only 21.4% of Tor nodes are exit nodes. Are we wasting this precious resource by running non-exit traffic through these nodes?
More important than "what percent of nodes are exits" is: what percentage of throughput is provided by exit nodes?
Based on my (probably wrong) hacking at the stats, exit nodes handle a bit over a third of the actual traffic in the Tor network, 36% in the last snapshot I have at hand.
Small nodes tend not to be exits, and exits tend to be quite large (100 Mbps or greater), so limiting exit nodes to only handling exit traffic would actually decrease the overall throughput of the network.
-andy
On Sat, 31 Aug 2013 09:58:12 -0700 Andy Isaacson adi@hexapodia.org wrote:
On Sat, Aug 31, 2013 at 11:32:55AM -0500, Pascal wrote:
http://torstatus.blutmagie.de indicates that only 21.4% of Tor nodes are exit nodes. Are we wasting this precious resource by running non-exit traffic through these nodes?
More important than "what percent of nodes are exits" is: what percentage of throughput is provided by exit nodes?
https://compass.torproject.org has some of these stats.
On Sat, Aug 31, 2013, at 05:32 PM, Pascal wrote:
http://torstatus.blutmagie.de indicates that only 21.4% of Tor nodes are exit nodes. Are we wasting this precious resource by running non-exit traffic through these nodes?
-Pascal
No. The non-exit traffic masks which relay the exit traffic is being handled for - which would be quite obvious most of the time.
31.08.2013 18:32, Pascal:
http://torstatus.blutmagie.de indicates that only 21.4% of Tor nodes are exit nodes. Are we wasting this precious resource by running non-exit traffic through these nodes?
Hi,
whenever or not these numbers are accurate or not, yes exit nodes transport all Tor related traffic.
They transport client traffic into the network (if they are guards) and back to the client, relay it to exits (where traffic exits) and exit-traffic to and from the destination.
The question is whenever or not it beneficial to change the role of nodes to achieve better ... what... performance... anonymity.
For example I thought myself if exits should be guards, because if I ran two exits which both got the guard flag and I don't set family on them there is some probability that traffic enters at one point (my exit no.1) and exits to my other exit.
Well Tor people are well educated and have more experience in that field.
Exits only transporting exit-traffic seems to be beneficial for performance, but to what expense on the anonymity side of things?
Best, Sebastian G.
tor-relays@lists.torproject.org
-
Andrew Lewman -
Andy Isaacson -
Geoff Down -
Pascal -
Sebastian G. <bastik.tor>