On Tuesday 23 July 2002 04:41, Oskar wrote:
> On Sun, Jul 21, 2002 at 11:55:20PM -0700, Ian Clarke wrote:
> <>
>
> > > 2) The overload prefiltering code discriminates against requests which
> > > are far from the node's estimated specialization when the node is
> > > overloaded, making them more likely to be QR rejected.
> >
> > Not to nitpick - but I have always been somewhat uncomfortable about
> > assuming a single area of specialization for a node.  In many
> > simulations I have seen, nodes will frequently specialize in two or more
> > areas.
>
> I thought the point of this code was that it used the relative density
> of the neighboring portion of the keyspace, thus not assuming anything
> specific about the specialization properties. That was certainly a
> requirement for my support.

I'm not sure what you mean.  The current code keeps track of the success rate 
of incoming requests as a function of key value (key space is divided into 
256 bins) and uses this estimate to decide whether to answer the request or 
not.

It doesn't try to pick a single winning peak.

--gj

-- 
Freesite
(0.4) freenet:SSK@npfV5XQijFkF6sXZvuO0o~kG4wEPAgM/homepage//

_______________________________________________
devl mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to