[e2e] Simple Question on TCP Window Size

David P. Reed dpreed at reed.com
Wed May 9 06:09:19 PDT 2001


At 12:04 PM 5/9/01 +1000, George Michaelson wrote:

>   > > Why don't we set the reduction ratio a litter higher?
>   >   This will only slow down the convergence, but has no effect on 
> fairness,
>   >  if
>   > the UDP flow does not respond to congestion.
>   >
>   >   Have a nice day.
>
>Doesn't that depend on the life of the flows? Or are any ameliorations
>irrespective of the duration of either competing TCP or UDP flows useless?

I think most UDP flows (as opposed to short duration exchanges like DNS) 
would benefit from using ECN and RED information.  Surely those who do 
these commercially would be interested in working on such adaptations.  For 
example, adjusting the coding rate of streaming video and audio can be 
responsive to congestion signals of this sort, and existing servers have 
the mechanisms to do so, just not the control loop.


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





More information about the end2end-interest mailing list