Clustering question

[delabu]

I have identical nodes that need to serve non-identical data.

Each node is a image server that connects to a large (>300GB) proprietary format image data file.

Some of the files that are used often will be duplicated accross two or more nodes and others won't.

Is there a way to set up the nodes so that they can be clustered, without every node having access to the same resources?

If the call to one node fails (because it may not have the right file), will the cluster automatically retry using a different node. Though, I don't think load balancing by exception will be so efficient.

Any suggestions?

[delabu]

Well, since no one jumped up and said, idiot, it's so easy to do that, it's either:

1. My question wasn't clear enought
2. No easy answer to it.

Or, maybe it's not even a deployment question . . .

Any ideas if I'm heading in totally the wrong direction (or am I just impatient. . . )

Any suggestions where else I can ask?