[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [ns] Link delay, what is it?



... but there is no feature to simulate processing delay, right?
Or is there? At least I never found out about one.

Therefore, in simulating a real network, I would at least use the
value obtained from a ping instead of the real propagation delay for
the link delay value in ns, to account for tcp/ip stack overhead.
Which is of course no good approximation as processing delay will
indeed vary depending on the packet size and probably the protocol
as well.

Cheers,
Michael

> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]]On Behalf Of
> Debojyoti Dutta
> Sent: Thursday, June 07, 2001 2:02 AM
> To: Huan Pham
> Cc: NS mailing list; Nguyen Huu Thanh
> Subject: Re: [ns] Link delay, what is it?
> 
> 
> When you set the link delay you set the propagation delay. The
> transmission delay is given by 1/BW and the queuing delay 
> depends on the
> buffer capacity, average queue length and the queueing discipline.
> 
> For longer packets, delay will be longer.
> 
> Hope this helps
> Debo
> 
> On Thu, 7 Jun 2001, Huan Pham wrote:
> 
> > 
> > Hi NS users,
> > 
> > My understand is that the Delay of a packet over a link 
> consists of 4
> > elements:
> > 
> > 1 - Queueing Delay
> > 2 - Processing Delay
> > 3 - Propogation Delay (negligible for most scenario, 
> excluding satelite)
> > 
> > 4 - Transmission Delay (= packetSize / Bandwidth)
> > 
> > My question is, in NS, when we set a delay for a link, what delay we
> > mean here. Of course it not the queueing delay, 
> transmission delay. So I
> > believe it must be a sum of the processing delay and the propogation
> > delay. IS THAT CORRECT? If so, is that value the same for 
> all packets
> > regardless of their packet sizes? It's questionable, as I 
> believe longer
> > packets may require more processing time.
> > 
> > Thanks for your reply.
> > 
> > 
>