Re: [ns] Can we run ns in parallel?

[John Heidemann <johnh@ISI.EDU>]

>	I would like to know if we can run ns on parallel processors.
>Is there anyway where we can parallelize the source sode so that it can be
>run on parallel processors simultaneously.

PDNS, mentioned by Lloyd and George (and linked to off the main ns
page), provides parallelism within a single experiment and is
recommended if that's what you need.

Many of the experiments I've seen don't exist in isolation, though.
Lots of folks run 100s of experiments (for replication or to test
different parameters).  If this is your case you should also consider
just running multiple independent instances of ns.  Although a single
job won't be any faster, the total load parallelises perfectly (to the
limit of experiment granularity).

Folks at RPI have done some work to support this (search for their
"experimental factorizaton" work), or you can just use a simple shell
script to run the jobs on different machines.

   -John Heidemann