[Pogamut-list] GRID status

[nvh]

Re: GRID status

Author: nvh

Hi,

I want to use GRID as well to speed my experiments up, but I don't quite understand how it works. I have a cluster server available at my institute, but I can only submit jobs that will distributed over a number of nodes. I don't have any control on which node the jobs will be running.
1) Can you explain a bit how the architecture works? 
2) As I understand it there is one driver to which the nodes connect, and you connect with a java client (the code) to the driver and submit your experimentDescriptors that will be distributed by the driver to the nodes. Is this correct?
3) Can you give a brief example how to setup the different parts and how they should be configured?

Thanks!

Niels