On Wed, 27 Jan 1999, Reiner Ludwig wrote:
> >>I do wish that TCP could figure out for itself how to avoid growing
> >>very long queues. As far as I know nobody has come up with a really
> >>good algorithm to do it in the absence of SQ, or preferably ECN.
> >
> >I guess without explicit feedback from the network it should be impossible
> >for TCP to figure that out.
>
> Before this starts off a debate that I think wouldn't belong here: yes, it
> is _not_ necessarily impossible. RTT increases can in theory be used as a
> signal for congestion.
RTT increases/variations may be due to :
- which packet you are timing (note that with the cumulative ack
policy, the information is far from reliable)
- ephemeral increase/variations of cross traffic
- delays at the operating system at the end-point
- delays due to the protocol (e.g, delay ack in TCP)
- route changes
Besides this, the information about the RTT reaches you a certain
delay, which makes it increasingly worthless when the following happens :
- propagation delay increases
- bandwidth at the bottleneck increases.
We are conducting a study on the traces collected by Vern Paxson
for his thesis to see if congestion may be "predicted". Results so far
give a negative answer.
Saad
___________________________________________________________________
Saad Biaz (Abou Youssef & Rim) Lecturer
Texas A&M University Office : (409) 845-5007
Department of Computer Science Home : (409) 862-9135
College Station, TX, 77843-3112 Fax : (409) 847-8578
___________________________________________________________________
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 09:12:20 EST