[tor-talk] hidden services and stream isolation

[Nick Mathewson]

On Fri, Sep 7, 2012 at 11:02 AM, Fabio Pietrosanti (naif)
<lists at infosecurity.ch> wrote:
> On 9/7/12 10:28 AM, Nick Mathewson wrote:
>> Hidden services are already isolated from each other, and from non-hs
>> client traffic.  The only additional feature for stream isolation to
>> provide would be isolating some streams sent to a hidden service from
>> another -- for example, so that if you have SocksPort 9050 and
>> SocksPort 9051 defined, any streams to a given HS over port 9050 will
>> be sent on a different circuit than streams over 9051.
>
> So, if i would do a piece of code that want to "benchmark" the
> performance of different full-circuit to a tor-hs, i may just:
> - setup two Socks port 9050,9051, 9052, 9053, 9054,9055

For certain values of two, sure.

> - create 5 connections using different socks port
> - exchange traffic to benchmark bandwidth and latency
> - pick-up the faster one

I was with you to this part.  I thought you were trying to benchmark
performance, not pick the fastest one at the expense of hosing the
network.

-- 
Nick