We have a 250-node Red Hat 7.2 cluster and we want
to break it down into 5 racks of 50, each with a head node. I want to make each
child node a diskless client of the head node. What is the best approach? I
noticed several HOWTO's on different methods. I just wanted to get some feedback
on the best method for purposes.
The cluster is mainly used for image analysis of
biotech data, the data being analyzed is mounted via NFS.
Thanks,
Chuck
|