building a linux experiment server

[catechu]

Hello everyone,

I want to build/buy a simple server for running data mining experiments (so it should be fast), storing files, and serving files to a few outside users. I would like it to be remote access via my wireless network (this may not be the standard term...), so that I do not need to purchase a monitor. My budget is $1000.

I have been out of touch with the hardware scene for a while, so what are your opinions on:

1. building from scratch vs. buying pre-made
2. relevant hardware specs (e.g. processor, dual-core)

Thanks for your help!

 

[tlarkin]

what kind of data mining and how will you transfer data to it? How many users will be using it? What services do you plan on running?

 

[catechu]

I'm planning just to run experiments locally, such as corpus analysis -- so no spidering or anything like that.

I am not sure what a good way of transferring data to and from the machine is. I was planning to use an external hard drive for the large files.

4 users maximum (given permission by explicit authentication).

Other than a web server, I will run no other services exposed to the outside world.

 

[tlarkin]

hard drive space, web access, and rysnc or sftp is all you would need to share the data amongst clients to give you a few ideas.