[tor] Re: more info needed to be able to help

  • From: Mark Smith <torop1@xxxxxxxx>
  • To: <torservers@xxxxxxxxxxxxx>
  • Date: Sun, 27 Feb 2011 15:43:41 -0500



> Date: Sun, 27 Feb 2011 17:23:17 +0100
> From: moritz@xxxxxxxxxxxxxx
> To: torservers@xxxxxxxxxxxxx
> Subject: [tor] Re: more info needed to be able to help
> 
> Hi,
> 
> Am 26.02.2011 20:29, schrieb tagnaq:
> > Yes, I was surprised that I didn't find any thread about that (severe)
> > problem (e.g. on tor-relays or here) (beside some side talk in a
> > different thread).
> > No explicit "This is the problem, how can one fix it?" - email.
> 
> http://archives.seul.org/or/relays/Feb-2011/msg00006.html


Could it simply be that the relays are offering more bandwidth than users need? 
That's what I always assumed. None of my relays use their full bandwidth, 
although they aren't exits either. Still, the exits may not be a limiting 
factor anymore.

According to Tor Metrics, the network has been carrying ~700 MB/s in recent 
weeks. Each circuit includes three nodes, so the actual amount of unique data 
is 1/3 of that**, or ~233 MB/s, which needs to find its way to an exit. I just 
pulled a CSV of relay information from Blutmagie and summed the actual (not 
advertised) bandwidth of the exits, which is 298 MB/s. So, they're already 
carrying more bandwidth than they need to in order to handle all the unique 
data (the reason it's more is because exits also carry middle and entry 
bandwidth).

** While the network carries 700 MB/s total, it's the same data at different 
nodes, at different times, so the exits only need to be able to handle 1/3 of 
the data, as long as they are 1/3 of the nodes on a circuit.
                                          

Other related posts: