[ns] slow start an congestion wondow question

[Dimitris Velenis]

Hi Sunesh,

Try commenting out the "$tcp1 set ssthresh_ 1" command.
Your agent won't go in Slow Start with so low a ssthresh value...

Dimitris.

On Thu, Apr 18, 2002 at 09:30:16AM +0300, Sunesh Kumra wrote:
> 
> Hi All,
> 
> I am trying to use the RFC793edu TCP agent and trace it's congestion window and
> 
> throughput under 2 cases 
> a) when using slow start
> b) when using additive increase
> 
> the problem is that I am not getting the expected results … I would expect that
> 
> the throughput would increase when I use slow start and also that the 
> congestion window builds up much faster in case of slow start ... but both 
> things do not happen ... the throughput decreases !! and CW remains the same. 
> Isn't that what should happen ? 
> 
> I have attached the section of the code where I am initializing the Agent :
> 
> # create a TCP-RFC793edu connection
> set tcp1 [$nssim create-connection TCP/RFC793edu $n0 TCPSink $n2 1]
> $tcp1 set fid_ 1
> $tcp1 set window_ $maxwnd
> $tcp1 set rtxcur_init_ $initialtimeout
> $tcp1 set add793slowstart_ true
> #$tcp1 set add793additiveinc_ true
> $tcp1 set packet_size_ $mpktsize
> $nssim at 0.0 "$tcp1 set ssthresh_ 1"
> 
> I just uncomment the line for additive increase and comment the line for 
> slowstart in other case.
> 
> and this is how I record the CW window :
> proc record {} {
>     global nssim tcp1 winfile maxwnd
>     #Sampling interval
>     set time 0.1
>  
>     #WINDOW FOR TCP1
>     #Read congestion window of tcp1
>     set curr_cwnd [$tcp1 set cwnd_]
>     #Compare the value of cwnd to maxwnd to decide actual sending window
>     if { $curr_cwnd < $maxwnd } {
>         set curr_wnd $curr_cwnd
>     } else {
>     set curr_wnd $maxwnd
>     }
>     #current time
>     set now [$nssim now]
>     # write the file as window vs time trace
>     puts $winfile "$now $curr_wnd"
>     #re-schedule procedure
>     $nssim at [expr $now+$time] "record"
> }
> 
> can anyone locate a mistake ? Thanks for your help in advance.
> 
> Regards,
> sune.sh
> 
> ----- End forwarded message -----

-- 
Dimitris Velenis                              <dbelen@netmode.ece.ntua.gr>
Network Management & Optimal Design Lab
National Technical University of Athens