onion_router stress-testing...

[Bruce Montrose <montrose@itd.nrl.navy.mil>]

See the reports for the following three tests in 
http://freehaven.net/~montrose/tests.zip

1) httpap with onion_router
2) privoxy with onion_router
3) privoxy without going through onion_router

In summary,

test#2 was only slightly worse than test#1 probably because privoxy does 
data scrubbing and httpap did not.

Both test#1 and test#2 perform horribly under great stress.

test#3 performed wonderfully! Unfortunately it means that the blame for the 
poor performance in tests #1 and #2 is either directly or indirectly 
related to use of the onion_router.

Can you think of any instrumentation of the onion_router code that might be 
used to give us information that might shed some light on what the hold up 
is? I could rerun test #2 with this instrumentation in place to gather the data.