Exit Balancing Patch

[Mike Perry]

Thus spake Michael_google gmail_Gersten (keybounce at gmail.com):

> My thoughts on entry guards:
> 
> #1. They should advertise a bandwidth of at least twice my incoming
> speed. Maybe slightly more. If their speed is any lower, then there's
> a good chance that I won't ever be able to get full speed.

There's a good chance you won't get full speed anyways. Full down
speed on my link is 6Mbit/sec. I can gaurantee you I will never see
that through Tor. It also does not mean I should be only selecting
bandwidth guard nodes..

> #2. The chance of their being chosen should be based on how much
> faster than my incoming bandwidth that they are, as well as their
> "overloaded" factor -- what percentage of the time are they maxed
> (either CPU maxed, or bandwidth maxed. I'd say memory/swapping limited
> if there was a way to tell). I'd rather get a lower capacity machine
> that is more than what I need if it is being ignored by the rest of
> the network.
>
> #3. For anonymity, I want an entry guard that is used by others. So I
> want an entry guard that not only serves me, but several others.
> Suddenly, I want a guard that has not just more than my speed, but
> more than the speed of the other users. If I'm the fastest client
> using that guard, the rule of "slightly more than twice what I need"
> applies; otherwise, the speed that the others are using should
> dominate, and I want to make sure that there is leftover bandwidth to
> support me.
> 
> Thoughts?

My thoughts are that we fix the current issues with as straightforward
and obviously correct fixes as possible, and eliminate voodoo magic and
unexplained, unprovable, and uninteligible formulae :)

Once this is done, we can go crazy with more complicated schemes. But
we need to fix these major issues first and propogate these changes to
everyone before we can even hope to do any fine-grained tweaking.

-- 
Mike Perry
Mad Computer Scientist
fscked.org evil labs
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.torproject.org/pipermail/tor-dev/attachments/20070727/7550fef5/attachment.pgp>