Re: [ledbat] draft-ietf-ledbat-congestion-01a.txt

Hello,

I cannot follow this statement:

> However, once the initial flow yields, the late comer immediately
>   measures the true base delay and the two flows operate from the same
>   (correct) inputs.

What we observed was that the decrease of the first flow and the increase of the second one resulted in a roughly constant amount of filling state in the modem's buffer. Even if the late comer eventually takes over the whole queue it was unable to correctly measure the base delay (self-measuring problem, same reason why the base delay gets wrong after 10 minutes free of cross-traffic).

If one extends this issue to more than 2 competing LEDBAT flows it becomes rather unlikely that the late comer will measure the correct base delay during the startup phase.

Joerg

> -----Original Message-----
> From: ledbat-bounces@ietf.org [mailto:ledbat-bounces@ietf.org] On
> Behalf Of Stanislav Shalunov
> Sent: Montag, 22. März 2010 22:50
> To: ledbat@ietf.org
> Subject: Re: [ledbat] draft-ietf-ledbat-congestion-01a.txt
> 
> The current draft version is now at
> http://tools.ietf.org/id/draft-ietf-ledbat-congestion-01.txt
> 
> 
> On Mon, Mar 22, 2010 at 3:55 AM, Stanislav Shalunov
> <shalunov@bittorrent.com> wrote:
> > draft-ietf-ledbat-congestion-01 missed the submission deadline, so
> > we'll will be discussing tentative textual changes in the
> presentation
> > tomorrow. However, since the text does exist, I thought I'd send it
> to
> > the mailing list -- attached.
> 
> 
> 
> --
> Stanislav Shalunov
> BitTorrent Inc
> shalunov@bittorrent.com
> 
> personal: http://shlang.com
> _______________________________________________
> ledbat mailing list
> ledbat@ietf.org
> https://www.ietf.org/mailman/listinfo/ledbat