[Noisebridge-discuss] New Years Resolution: keep the Internet safe for free speech?

Andy Isaacson adi at hexapodia.org
Tue Jan 3 07:05:30 UTC 2012


On Tue, Jan 03, 2012 at 12:46:23AM +0100, John Adams wrote:
> Not only that, by having disparate nodes, it reduces the
> single-point-of-failure issue by having all the monies in a single
> node.

To address this point a little bit, we can explore
http://torstatus.blutmagie.de/ and specifically the CSV export at
http://torstatus.blutmagie.de/query_export.php/Tor_query_EXPORT.csv

Currently torstatus is showing 2416 relays.  Of those, 768 are exit
nodes, but a bunch in the long tail seem to have their bandwidth limit
set to 0 so there are only 718 with non-zero bandwidth numbers.

Adding up the total bandwidth reported by torstatus (which counts in
kilobytes/second, but I'll convert to MB/s for convenience), we see that
noisetor adds up to 40 MB/s.  The total bandwidth of the entire network
adds up to 1,328 MB/s.

However, this counts relays (middle nodes) as well as exit nodes.
Looking just at exit nodes, torstatus reports 518 MB/s.

The distribution is heavily weighted; 23 nodes provide half of the
reported exit bandwidth and 92 nodes provide 90% of the exit bandwidth;
the remaining 620-odd exit nodes provide 10% of the exit bandwidth.

According to this snapshot, noisetor is providing about 8% of the exit
bandwidth on the Tor network.  torservers.net is around 12.4%,
formlessnetworking.net around 7.5%, ipredator.se around 7%, torland.me
around 6.6%, UVT.nl around 6.3%, IDXF.com around 4.8%, CCC.de around
3.6%, and william.ir around 3%.

Together those 9 organizations provide 59% of the Tor network's exit
bandwidth.

I would be comfortable with noisetor doubling our throughput based on
additional donations.  We shouldn't grow too far faster than the rest of
the Tor network, though -- no single organization should control a
significant fraction of the network, and 10% of the exit bandwidth seems
like a reasonable "do not exceed" line.

Hopefully other people or organizations will step up to add additional
exit bandwidth to the Tor network, so that we can grow noisetor past
1Gbps safely.

Again, thanks for asking the questions that lead to insightful data
analysis like this!

-andy

My horrifying awk script to compute the above, preserved for posterity:

curl http://torstatus.blutmagie.de/query_export.php/Tor_query_EXPORT.csv | awk -F, '{printf("%-6s %20s %2s %4s %3s %s\n",$3,$1,$2,$10==1?"exit":"",$20==1?"BAD":"",$6);if($10==1){e+=$3}t+=$3}END{print t" "e" "NR}' | sort -nr | awk '/exit/{e+=$1;p=1}{printf("%s %s\n",$0,p?e:"");p=0}' | cat -n | less



More information about the Noisebridge-discuss mailing list