[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: performance tuning...

On Thu, Apr 14, 2005 at 02:55:49AM -0400, Roger Dingledine wrote:

:No, I think the limit you're running into is that there isn't much more
:traffic that needs to be carried.

Ah, I hadn't even considered that possibility!

:high bandwidth servers will be able to attract more use. (The tradeoff
:is that I don't want somebody to advertise an incredibly huge bandwidth
:and attract most of the circuits -- so I'm slowly raising the cap as
:the overall network capacity grows.)

Makes sense to keep this bounded in some way.  For now the hardcoded
cap seems fine, looking forward to the pie in the sky future perhaps
this could be cased on some run time hueristic like N standard
deviations above mean advertized bandwidth (I don't have a clue what's
a resonable value for N).

This is how we handled storage quotas at the old AI lab, storage isn't
network but they both have a certain gaseous nature...

:Well, you could always run spoon2 on the same machine. :)
:(If you do, please set the MyFamily torrc directive.)

I've been thinking of setting up a second system (had Borg0, but I
broke the OS hard, not quite sure how...)