[altq 1481] Re: RED implementation

[Nguyen-Tuong Long Le <le@cs.unc.edu>]

On Sun, 12 May 2002, Nicolas Christin wrote:

> On Mon, 13 May 2002, Kenjiro Cho wrote:
> 
> > Nguyen-Tuong Long Le wrote:
> > > I can follow and understand the code but my question is
> > > why does ALTQ/RED compute (prob/(2 - count*prob)) instead
> > > of (prob/(1 - count*prob)) as in the Floyd and Jacobson paper.
> >
> > It was recommended by Sally and implemented in ns.
> > I don't remember where it was described, though.
> 
> 	Correct me if I'm wrong, but isn't that a way to implement the
> gentle_ variant of RED? It was never published, but Sally Floyd has an
> unpublished note about it on her website. The basic idea of gentle_ RED is
> to drop with proba max_p at max_th (as in RED), and instead of jumping
> directly to 1 at max_th, the drop probability increases linearly to reach
> 1 at 2*max_th.

What you are saying about gentle RED is correct. However, I
was asking why the slope of the drop function is different from
the one in the original paper when minth <= avg <= maxth.

ALTQ doesn't have gentle RED yet and I'am currently implementing
it (actually, that's why I asked about the slope since I
want to continue on with gentle RED where the drop function
of vanilla RED left off). I'd wanted to do it last summer but
I got sidetracked and decided to do something else.