[vox-tech] Clustering two Linux machines

[Ken Bloom]

Matin Hashemi wrote:
> Hello everybody :-)
>
> I would like to know how to make two Linux machines work as one
> computer. I think it's called "clustering" and a quick Google search
> showed me that there are quite a few different programs and Linux
> distributions for this purpose. But my problem is that I can't start
> everything from scratch because we already have one Linux machine up
> and running and everybody in the lab is using it. I just want to add
> another computer to that Linux machine and make it more powerful. We
> can't install OS on the first computer but we can do whatever we want
> with the new one.
>
> Thanks a lot for your help
> Matin
Can you tell me what cluster distribution you'd like to use, and what
features you hope to gain?

OpenMosix, which provides transparent process migration, is no longer
maintained and never became stable on the Linux 2.6 kernel.
(http://en.wikipedia.org/wiki/OpenMosix)

Beowulf requires explicit parallel programming, AIUI, and is not
suitable for automatic load balancing.
(http://www.phy.duke.edu/~rgb/Beowulf/beowulf_book/beowulf_book/node9.html)

I don't think there's any clustering technology in the mainline Linux
Kernel.

You should consider a solution with NFS, NIS, (these can be used to make
the two systems look roughly the same), and some sort of distributed
batch processing software (if your users have that kind of workload).
Explicit parallel programs can be written for this kind of setup, using
mpich (and possibly other MPI implementations).

--Ken