[e2e] Congestion control and avoidance - QoS?

Michael Welzl michael at tk.uni-linz.ac.at
Fri Aug 17 03:00:37 PDT 2001


> Can I view congestion control and avoidance as one of the QoS
> mechanisms? If so, I assume the Internet has evolved from best-effort
> service to "congestion controlled and avoided best-effort
> service" (assuming all end-systems are deployed with TCP-friendly
> algorithms) ???
> What was the reason that TCP was chosen to control and avoid 
> congestion
> whereas UDP was left to be free from congestion control and 
> avoidance then
> TCP-friendliness became an issue? If the answer to above 
> question is yes,
> then I assume current Internet offers two types of service: 
> one of them is
> "TCP-friendly throughput guarated best-effort service" and 
> the other one
> is "better-than TCP-friendly throughput guaranteed 
> best-effort service".

This point of view explains why I personally don't believe that
TCP congestion control would "make it" in the commercial world
nowadays. That is, if everybody had their TCP without congestion
control, nobody would bother restricting his or her own bandwidth
("what is the commercial gain for the end user, anyway?") and
we would all be stuck.

Or maybe it would be implemented in Windows and "make the Internet
faster". That might even work. Hmmm ... which makes me think:
Could Microsoft actually "make the Internet faster" for us all
by implementing some new and clever congestion control mechanism?

I believe TCP SACK and ECN are supported by new Windows versions,
aren't they? Is there already any measurable gain for the Internet
on a global scale?


More information about the end2end-interest mailing list