[ns] Re: ns/sack/loss strange behavior

[Padmaparna Haldar <haldar@ISI.EDU>]

The SACK implementation of full-tcp is currently broken and is being
worked on by Kevin Fall. so you might want to check out the ns daily
snapshot once kevin is done fixing.

--Padma


On Fri, 3 Aug 2001, Tom Dunigan wrote:


> I'm seeing some unusual behavior from ns 2.1b8a  (if you're not
> the right person to share this with let me know)
> Attached is a simple sackloss.tcl that runs SACK with a given
> loss rate.  Instead of nicy steady decline in throughput as the
> lossrate increasess, this SACK version (newreno test had no such anomaly)
> has strange behvior at .01 loss rate, summary from a sequence of tests:
>
> loss   bandwidth
> .0001 6184
> .002 3316
> .005 2064
> .01 131422
> .02 913
> .05 317
>
> my ns is running on a linux box
>
> thanks for any pointers
>
>

-- 
------------------------------------------------
Be true to your work, your word, and your friend.
--Thoreau

Padmaparna Haldar