[ns] queueing mechanism in ns
Fri, 28 Jul 2000 16:28:46 +0530 (IST)
even if queue is empty it takes finite time to transmit a packet (in my
case .6ms) because the link has a certain finite capacity. this should
appear as the difference in time between the time a packet arrives at an
empty queue and the time it is served out by the server ( i.e, dequeued).
hope i am not garbling up fundamentals.
On Fri, 28 Jul 2000, Bilgin Metin wrote:
> I think the last column gives packet id. Thus, in trace file we should
> compare two lines which is beginning with different sign "+" and "-" but
> which has the same packet id. In the beginning of a simulation , because
> the queue is empty, the queuing delay is zero and the line with the same
> packet id is next to other one. . Usually after 3 second later from
> beginning of a simulation, there are a lot of lines of other packets
> between the lines with the same packet id. Hence, delay will not probably
> Bilgin Metin
> ----- Original Message -----
> From: "Punyaslok Purkayastha" <firstname.lastname@example.org>
> To: <email@example.com>
> Cc: <ns-users@ISI.EDU>
> Sent: Thursday, July 27, 2000 11:47 PM
> Subject: [ns] queueing mechanism in ns
> > hi,
> > i am not quite sure i understand the exact queueing and delay mechanisms
> > of ns. i am employing a tcp source going through a
> > single link. the first two lines of output in my trace file are
> > + 0 0 1 tcp 750 ------- 1 0.0 1.0 0 0
> > - 0 0 1 tcp 750 ------- 1 0.0 1.0 0 0
> > now my tcp packets should suffer a transmission delay of .6 ms. however
> > the output shows that the packet suffers no delay at all ( it gets
> > enqueued and dequeued at the same time).
> > however note (at a later portion of the trace file)
> > r 0.0206 0 1 tcp 750 ------- 1 0.0 1.0 0 0
> > the propagation delay of my forward direction link is 20ms; so that i do
> > get that the said tcp packet arrives at the receiving end 20ms+.6ms later.
> > i am perplexed by what is going on.
> > Punyaslok.