Re: about ns-2

[bmah@ca.sandia.gov (Bruce A. Mah)]

Dapeng Wu writes:

> > Yes, I'd agree with Kathie.  If the intention is to look primarily at the
> > ATM layer (with less emphasis on IP), there are some other simulators,
> > (e.g. ftp://isdn.ncsl.nist.gov/atm-sim/ or
> > http://HTTP.CS.Berkeley.EDU/~bmah/Software/Insane/).
> > 
> > - K
> 
> I did have NIST ATM Network Simulator.  It's a powerful simulation for
> ATM network.  But its function in simulating TCP/IP traffic is very
> limited compared with ns-1 and ns-2.  
> 
> What I'm concerned with is a hybrid IP/ATM networks since the
> integration of IP router and ATM switches is a very important research
> area right now.  I'm not sure if ns-2 developers have  interest in
> adding ATM stuff to it so that it can be more powerful and useful.

Dapeng--

INSANE was designed to look at some performance issues in IP-over-ATM networks 
for my dissertation.  The model in this simulator is that ATM is a datalink 
layer used by IP.

INSANE has a (IMHO) a fairly good cell-level transport, but it's not pure 
Q.2931 by any means.  It does have a pretty good set of Internet source 
traffic models (a combination of tcplib models and some of my own devising), 
but it doesn't support IP multicast or the wide variety of TCP models that 
ns-2 does.  It might be possible to take some of INSANE's cell-level 
functionality and put it into ns, or vice versa, but I haven't given this 
matter much thought (yet).

Bruce.

PGP signature