On Fri, 23 Mar 2001 10:56:10 +0800, Wang Hui said:
I am about to set up a small Network Simulation lab as to do some
network protocol simulations.
Congratulations, and good luck.
The first thing to do is to choose the software and its platform.
No, the *first* thing to do is understand *what* you're trying to simulate
and *why*.
Are you trying to understand the protocol at all? You'd probably be
better off spending the time and effort reading a good book on the
subject, or doing a literature search for research papers already done.
Are you trying to understand a *specific aspect* of a *specific protocol*?
For instance, "failure modes of the SMTP protocol when communicating with
a node that implements the protocol improperly" (I've studied *that* enough
myself)? Or are you trying to understand oscillation modes in BGP flaps
(which I don't understand well, except as they apply to queue management
for an SMTP server - when the routes are flapping, you start piling up
queues).
If so, you need to specify what protocol, and what aspect,
you're studying. The question you are asking will make a difference in the
tools you will need to answer it.
But as for me, I have no experience of working on it, so I have no idea
of that. Anyone has any idea?
Idea: Get up to speed on it before trying to do simulations. Or at least
figure out something more specific than "I've been told to set up a lab".
Starting out without a clear goal in mind guarantees failure. Among other
things, lack of a clear goal means you can't tell when you've suceeded.
............................... Which network software is popular among
this field? Could you drop me a message? Or give me some web links?
The network software is *totally* *irrelevant* until you understand
what you're trying to do.
Valdis Kletnieks
Operating Systems Analyst
Virginia Tech