Best way to cluster Linux
 

Hey,

I am looking to build a really big server.

Since I don't have the financial resources to buy a big 10,000$ server, I was thinking of buying 4 250$ 1.7ghz/512mb computers.

I am trying to find the best way to link the machines together to make a single server.

I found the page http://72.14.205.104/search?q=cache:...ient=firefox-a
which is similar to the setup I want to do.

I will have a computer that will be a media centre that should be able to fetch the files from the cluster, as well I will have a mysql database on the cluster that will be used, amongst other things, to backend the media centre.

The above mentioned article talks about writing my own software to handle the cluster, but I don't really have the time for this.

Can anyone recommend any other articles that might help me set up this cluster?

David

Hey,
Last month I built a 7 node cluster (1 head node, six computing nodes) for doing numerical simulations. From my experience, the world of linux clustering is huge and there is great open source software out there and my main difficulty has been trying to choose amongst them to find the best ones for what I need to do. However, maybe my needs may be quite different to your's. I just want a big number cruncher, i.e a high-performance computer, and you sound like you need a high-availability computer.

As for literature, I am still a book-addict and find it easier to learn from books than from small articles or posts online. I got started on my way by reading an O'reilly book "High Performance Linux Clusters with OSCAR, Rocks, OpenMosix, and MPI" by Joseph D. Sloan. It was perfect for me. For building a server-cluster it may not be as relevant. On the other hand, I have heard good things about the book "The Linux Enterprise Cluster: Build a Highly Available Cluster with Commodity Hardware and Free Software" by Karl Kopper, No Starch Press. Maybe this book would be good for you.

You also may find useful articles here -> http://lcic.org/

good luck
mark

Beowulf network clusters.
 

Hi,

Look at Beowulf network clusters. The scientific community uses this for number crunching. Look at www.beowulf.org and Building a Beowulf System (http://www.cacr.caltech.edu/beowulf/.../building.html)

{
What is a Beowulf Cluster?
Essentially, any group of Linux machines dedicated to a single purpose can be called a cluster. To be called a Beowulf cluster, all you need is a central node that does some coordination (Ie, technically an IP Masq'ed network fits loosely into the Beowulf definition given by the beowulf mailing list.)

1.2 What can a Beowulf Cluster do?
A beowulf cluster is capable of many things that are applicable in areas ranging from data mining to reasearch in physics and chemistry, all the way to the movie industry. Essentially anything that can perform several semi jobs concurrently can benifit from running on a Beowulf Cluster. There are two classes of these parallel programs.
}

excerpt from Building Linux Beowulf Clusters (fscked.org/writings/clusters/cluster.html)

Rick.