Re: [ns] ECN!!!

[Kaleelazhicathu R R Kumar <kaleelaz@comp.nus.edu.sg>]

hi,
  thanks very much for the reply.
    Renjish.

On Thu, 26 Oct 2000, Sally Floyd wrote:

> > thanks for the reply.Can you please let me know if the two way tcp
> >supports ECN??according to the ns manual,no validations have been done for
> >this.I am using a tcp-full.
> 
> There is a validation test for ecn with two-way TCP, "./test-all-ecn-full",
> but it is not yet finished, and has not been added to the main
> validation test suite, because two-way TCP does not yet pass it
> correctly.  In particular, the test ecn_timeout_reno_full fails,
> because with two-way TCP, the congestion window is reduced twice
> when ECN is followed, in the same window of data, by a retransmit
> timeout.
> 
> The test ecn_smallwinEcn_reno_full also fails, because two-way
> TCP does not respond correctly to ECN when the congestion window
> consists of only one packet.  This is described in the cvs log for
> the file test-suite-ecn-full.tcl, at
> "http://www.isi.edu/cgi-bin/nsnam/cvsweb/ns-2/tcl/test/test-suite-ecn-full.tcl".
> 
> If anyone wanted to make the appropriate changes to the two-way 
> TCP implementation, I would be happy to check the changes into NS.
> 
> - Sally
> --------------------------------
> http://www.aciri.org/floyd/
> --------------------------------
>