monitoring delay per flow

[Konstantinos Psounis <kpsounis@cisco.com>]

Hi all,

I am monitoring the delay per flow using (say) the following code (the code
is not critical to my question, I am just presenting it here to avoid
confusion on what I do):
set l01 [$ns link $n0 $n1]
set fmon [$ns makeflowmon Fid]
$ns attach-fmon $l01 $fmon
set fcl [$fmon classifier]
set flow [$fcl lookup auto 0 0 $i]
set dsamp [$flow get-delay-samples]
set meandelay($i) [$dsamp mean]

I guess the "mean" delay above refers to the "queueing delay" +
"transmission delay" of the packets from node $no to node $n1.

When I am changing the delay (I guess the "propagation delay" of the link)
in the following code:
$ns duplex-link $n0 $n1 8MB 10ms DRR
I am getting differences in the "mean" delay from the monitoring object
above. The differences are not just the addition or subtraction of the
different propagation delays that I am using for the link. In other words if
I use 10ms and then 0ms, the "mean" delays I get from the monitoring object
do not differ by the 10ms -which should be the case if propagation delay is
also added to the delay samples of the monitoring object- and are not the
same -which should be the case if propagation delay is not added.

Am I missing something in the basic NS model of what a monitoring object
does, and what each delay is?

Thank you very much,

PS since I have a duplex link I am not also sure what the monitoring object
does in terms of whether it monitors only the packets of the simple link
from $n0 to $n1 or it monitors the packets of both directions. If the latter
is happening how can I achieve the former?
--
**********************************
Konstantinos Psounis
Summer Intern
Cisco Systems
IP Switching (ATM platforms) group
285 West Tasman Dr.
(408)5252542
**********************************