> [process migration, et al]
>
> I've done this before. You don't really mean process migration, you mean
> remote execution unless I misunderstood. Migration is something that you
> do to an existing process, remote execution is something you can do to
> a new process. The former is (a) a bad idea, and (b) extremely difficult and
> invasive to the kernel to fully implement. Fortunately, remote execution
> is extremely useful and quite easy to implement. In fact, I've just started
> rewriting a version of this I had running in Sun about 6 years ago.
Fabio clearly wrote about moving a running process from one machine to
another, so I assume my hint is of no value to him but still: If you
only want CPU bound processes executed remotely over the net, you can
always look at MPI (message passing interface) and if my fading memory
serves me right, there's a free implementation called Mpich. I once
compiled it on my Linux box and got the 'hello world' to compile but
that was just for fun since I have only one computer anyway.
MPI is pretty easy to use for your own distributed programs but cannot
cooperate with existing binaries.
Regards,
-- Eberhard Burr finger uhl4@rzstud2.rz.uni-karlsruhe.de for PGP key #include <stddisc.h> ------ electric cookie: "Marriage is like a cage; one sees the birds outside desperate to get in, and those inside desperate to get out." -- Montaigne