[e2e] Can feedback be generated more fast in ECN?

[David P. Reed]

I was being a bit too general.

At 05:00 PM 2/14/01 -0500, J. Noel Chiappa wrote:
>     > From: "David P. Reed" <dpreed at reed.com>
>
>     > the big elephant here is that looking at the packet level in a local
>     > router will never be the right place or time to solve congestion.  At
>     > best it is a way to help the network 's users cooperate to survive
>     > unforseeable transients that should be prevented earlier.
>
>Whoa, I was with you until you got to this. Are you saying thay you don't
>think there's any role for closed-loop congestion control (i.e. feedback based
>on actual congestion)? Or are you saying that intermediate routers should
>"know" more about their traffic, and monitor them for liekly future trouble,
>and not do congestion response on a packet-by-packet basis? Or something
>else?

You don't want to operate a network in a highly congested mode.  Router 
congestion should be rare, or your provisioning, your "admission control" 
(including such trivial things as restricting edge port capacities to 
reduce overcommitment of backbone capacity), or your closed-loop flow 
controls are not doing their job.

>And can you please explain what's so bad about classic closed-loop, especially
>in a network with such dynamic loads?

Nothing's *bad* about closed-loop control, IMO.  But congestion can result 
by interaction among independent loops, and interactions can be unstable.- 
David

--------------------------------------------
WWW Page: http://www.reed.com/dpr.html