Ok, let me explain the toplogy here. I have 4 switches, each switch is
connected with four nodes (which are 'upstream') and a root switch which is
also an 'upstream' node for other switches. Finally, this root switch is
connected with a single node (which acts as a sink). I have tried to draw it
here:
[H1]
^
|
|
[Root Switch]
^
|
|
[switch1] (this island is replicated three more
times and is connected with the root switch)
^ ^ ^ ^
| | | |
| | | |
[H2] [H3] [H4] [H5]
The hosts at the lowest levels are used to send traffic to the host at the top.
You can also find attached the trace and the tcl files.
In
addition to already mentioned problem, I also want to measure the
bandwidth consumed by the traffic at [Root switch] to [Host1] link. I
have some code in there but it isn't working right now. Let me know how
to get it working.
Thanks,
Salman
> From: svana...@gmail.com
> Date: Wed, 19 Dec 2012 23:10:45 +0000
> Subject: Re: [ns] Fixed Flow_id's ?
> To: salma...@live.com
> CC: ns-users@isi.edu
>
> Mate,
>
> No inkling what you are trying to say here.
>
> 1. Create a simplified scenario to reproduce the problem.
>
> 2. Share the TCL and trace file.
>
> Use it to explain your issue and steps you have taken to debug the problem.
>
> rgds
> Saravanan K
>
> On Wed, Dec 19, 2012 at 9:27 PM, Salman Malik <salma...@live.com> wrote:
> >
> > Hi,
> >
> > I have a topology where 20 hosts are connected and talking to another host
> > through TCP i.e. the first 20 hosts are attached to TCP source : [new
> > Agent/TCP] with an exponential application running on top : [new
> > Application/Traffic/Exponential], whereas the last host is simply the
> > TCPsink node : [new Agent/TCPSink].
> >
> > When I run the simulation, I see that the flow_id for all communications is
> > fixed to '0'. Also the packet size for the exponential sending application
> > doesn't match the value that I set it to e.g. I had packetSize_ set to 128
> > and I see packet sizes of 40 (this may be the ack bytes) and 1040 (which I
> > am not sure of) in the trace file.
> >
> > Comments please ?
> >
> > Regards,
> > Salman
