I've been interested for a long time in parallel programming with linux clusters and all that jazz... but how does one study cluster programming? This may seem like a silly question, as the answer is obviously "practice!", but if you think about it there are several issues:

a) I obviously don't have access to a large cluster of linux machines. I have my desktop and laptop, which probably isn't enough to get an idea of scaling/performance issues which are encountered in more realistic clusters. Are there "cluster simulators" or something?

b) which language/technology should I study? there seems to be thousands of different of different combinations, ranging from fortran, c, and java to lisp, zpl, erlang, and a whole host of others that you've never heard of.

c) which distro? a 30 second google search yields rocks, rhas, parallel knoppix, mandrake, etc.

Google "linux clustering" and you will end up with any number of "How To" sites and articles.

It's a fascinating topic and your questions are really more than can be addressed with a simple answer in this forum. As a result of your question I may have caught the bug to do more research on this topic myself.

I'm just about to learn about clustering too I've setup 5 virtual linux servers (using vmware) to see if I can get them all talking nicely together.