[tor-relays] weighting algorithm

[Mike Perry]

Thus spake Karsten Loesing (karsten.loesing at gmx.net):

> On Sat, Mar 12, 2011 at 04:57:58PM -0500, Flamsmark wrote:
> > Have you considered adding more bandwidth scanners? Is the quorum of 3/4
> > designed to prevent a pair of scanners from being evil, or to ensure that
> > representative sample is used?
> 
> My guess is that we should add more bandwidth scanners.  The main problem
> is that the bandwidth scanners are quite resource-intensive in terms of
> bandwidth (surprise!), RAM, and personal care by the operator.  Also, the
> bandwidth scanner results are tied to a directory authority vote, so the
> scanner should be run by the directory authority operators themselves or
> by a person trusted by them.
> 
> The requirement of having at least 3 bandwidth scanners running has to do
> with preventing a single bandwidth scanner from influencing the median.
> Even if there were 5 or 6 scanners, we'd require at least 3 scanner
> results before falling back to the relays' self-reported bandwidth.

Yes. We definitely want 5 or 6, but there will be a substantial
upgrade to the bw scanner code before we want to do this, I think.

There's a bunch of things that need to improve, most importantly their
stability and memory usage, but also their compatibility with the
latest sqlalchemy version, rather than some ancient one (or perhaps in
addition).

We can technically add them before these code updates, but it should
be a bit smoother to do it afterwords.



-- 
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-relays/attachments/20110316/0205126b/attachment.pgp>