[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ns] werid delay between enqueue and dequeue
Ok, thanks for your hints. I figured out the problem. In order to remove the
queue delay
to simulate circuit switching, I just simply replaced "queue" object with
"connector",
and disable delaylink to schedule the next dequeue time.
//s.schedule(h, &intr_, txt);
Then, each packet's enqueue and dequeue (they shouldn't exist) time are exactly
same.
Bo
Tarik Alj wrote:
> Bo,
>
> >Date: Thu, 6 Jul 2000 19:25:05 +0100 (BST)
> >From: Lloyd Wood <[email protected]>
> >X-Sender: [email protected]
> >To: Bo Wen <[email protected]>
> >cc: Tarik Alj <[email protected]>, [email protected]
> >Subject: Re: [ns] werid delay between enqueue and dequeue
> >X-url: http://www.ee.surrey.ac.uk/Personal/L.Wood/
> >X-no-archive: yes
> >MIME-Version: 1.0
> >
> >On Thu, 6 Jul 2000, Bo Wen wrote:
> >
> >> Thanks for your hints. But I checked out the queue code, that it seems there
> is
> >> no scheduled time related code. You know, ns uses the relative time, so how
> >> is this delay implemented? Since I am simulating the circuit switching,which
> >> assumes that there is no extra delay on the link except the link propagation
> >> delay,
> >> How can I remove this sort of queuing delay?
>
> what do you mean by "scheduled time"? Queueing delay is not a fixed or
> scheduled, it only depends on the number of packets present in the queue when a
> packet arrives. If you don't want to have any delay at all, set the queue size
> to 1 packet. But then expect some drops where you had delay...
>
> >
> >Do you have different link bandwidths?
>
> if circuit switching I assume the network is properly dimensionned... so there
> should be no bottle-necks, therefore minimum drops. But why would different link
> bandwidth would pose a problem?
>
> >
> >L.
> >
> ><[email protected]>PGP<http://www.ee.surrey.ac.uk/Personal/L.Wood/>
>
> Tarik