Pelican Linux Cluster.

Clusters are computers that work together to work for a common cause.  There are two basic types of clusters. One is a HA (high availability cluster) and the second is an HP (high performance cluster). Ha cluster allows several machines to be a failsafe for one program such as a web server that can not have down time. The HP cluster allows may programs doing the same thing to act as one computer. Science or math intensive calculations are usually done on these machines. These are sometimes know as parallel systems. What we will look at here is the High performance cluster.

Special programming is usually required on the HP clusters, traditionally single computer applications will not work usually on these systems in parallel mode.


More info: http://idea.uab.es/mcreel/PelicanHPC/
See also: http://www.hpc.lsu.edu/help/pelicannewusersguide.php

Step 1: Setup in a Virtual World.

Two collected videos to give you the lay of the land.



Another video:

Step 2: Actual Setup.

Doing a real setup: (Another collected video, but old but still applicable).

Step 3: Use Diskless Machines As Cluster Clients.

You can even use older machines as nodes. build them yourself from no longer used machines.  (or even new ones)

https://www.instructables.com/id/PC-personal-computer-into-a-sort-of-thin-client/

Step 4: Writing Software for Clusters.

Linux has some easily attainable software:

$ sudo apt-get install openmpi lam4-dev

$ mpicc mptest.c -o mptest
$ ./mptest

Also see:

http://www.linux-mag.com/id/5759/

http://www.math.umbc.edu/~gobbert/mpi.html

http://people.sc.fsu.edu/~jburkardt/c_src/mpi/mpi.html

Step 5: Other Mpi Clusters:

There are other clusters available:

https://en.wikipedia.org/wiki/Rocks_Cluster_Distribution

http://www.beowulf.org/
http://byobu.info/wiki/Building_a_simple_Beowulf_Like_Cluster_with_Ubuntu
http://devel.cs.stolaf.edu/projects/bw/wiki.real/index.php/Beowulf_Installation_Guide_2011_%28Ubuntu_Server_11.04%29

http://www.linuxhpc.org/pages.php?page=Cluster_Software