Re: [ns] TCP sequence number

[Qingjuan Gu <qgu@seas.upenn.edu>]

> Sequence numbers are signed 32 bit ints, so you've got 2G worth of
> packets before you start getting into strange numbers for sequence
> numbers.  Are you going to have a single flow send more than 2 billion
> packets in your simulations?
>


Thank you very much for your reply. I run my simulation only 10-15ms, so I
am not going to send a single flow more than 2 billion packets.

Best wishes.

Qingjuan

On Thu, 2 Aug 2001, Brian Lee Bowers wrote:

> On Thursday 02 August 2001 12:36, Qingjuan Gu wrote:
> > Dear all,
> >
> > I want to calculate the duplicated packets in tcp through
> > sequence number. but I am wondering if the sequence number is
> > wrapped round. Does anybody can explain this question
>
> The basic TCP implementations (Agent/TCP, Agent/TCP/Reno,
> Agent/Newreno, and Agent/Vegas) simplify sequence numbers in the
> following two ways:
> 	all connections start their sequence numbers at 0
> 	sequence numbers count packets, not bytes
>
> (I don't work with Full TCP, so I'm not qualified to say
> anything about how it implements sequence numbers).
>
> Sequence numbers are signed 32 bit ints, so you've got 2G worth of
> packets before you start getting into strange numbers for sequence
> numbers.  Are you going to have a single flow send more than 2
> billion packets in your simulations?
>
> --
> Brian Lee Bowers	|	RADIANT Team (Summer Intern)
> bowersb@lanl.gov	|	Los Alamos National Laboratory
>