Re: [RFC] Mass rDNS performance tweak

[David Fifield <david () bamsoftware com>]

On Fri, Jan 23, 2009 at 12:05:42AM +0000, jah wrote:
> On 22/01/2009 19:01, David Fifield wrote:
> > Likewise, what does this mean?
> >     capacity -= (float) drop_ratio
> > More drops means a greater reduction in capacity, everything else being
> > equal. But shouldn't drop ratios (defined as in the patch) of 6 / 5 and
> > 600 / 500 cause different absolute changes in capacity? In the first
> > case it seems we should reduce capacity by about 6; in the second, by
> > about 600. The code in the patch will change it by -1.2 in both cases.
>
> Remember also that we get a capacity decrease for each timed-out request
> so the overall decrease in capacity is much greater after 600 timeouts
> than after 6.

Ah, you're right of course. And me just having explained how that
identical effect affects congestion window growth.

David Fifield

_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://SecLists.Org