Beowulf cluster...?

Question

http://en.wikipedia.org/wiki/Beowulf_cluster

Since I figured out that my major was going to be computer programming rather than networking or anything else, I've ALWAYS been curious about something. I've always wanted to be able to set up a Beowulf Cluster, but I've never had enough machines to do it. I now have some old junk machines at the house with like 8 hard drives in them total, and since the interface for the hard drives is IDE and not SATA, I don’t really want to get an adapter, ect.

Anyway can someone tell me what their experience was setting up a Beowulf cluster? What amount of experience with Linux/Unix should you have (I currently have none), what are some good home applications? Such as creating your own cheap stand alone server, streaming media, ect.

Hari R, thank you for a very good answer.

Answer 1

Thats a wonderful Idea. Check whether you will be able to install a standard version of linux on them. Then try connecting to them with normal networking. You should have an idea of ip address , netmask etc.,
If all the machines have the same type of network card or for that matter , the drivers to the network cards are available in the standard linux, then you may not have much problem.
After that is sorted out, You have nothing more to worry. Clustering is something which takes little effort to set up.
The RAM should be sufficient on all the machines . Average of 128 MB would be good.
If you had 64 MB RAM , you may sometime land up waiting for a long time when that node is installed with the cluster image. Again there is something called file system which gets loaded on the RAM, which again would be slowing down processes.
I would suggest you first try understanding the Operating System and its concepts , doing something with small clusters on high availability, failover etc and scale up to Beowulf clusters.
ALL THE BEST.
You would need to know the following in OS concepts
1. File system 2. basic networking 3. tcp/ip 4. services 5. basic user management 6. installation methodologies.