* mount and format drives in cluster init * add monitor node array * disks are no longer persistent * now you can deallocate the entire cluster, and restart (disks, hostname) * a dedicated raid array for metadata
Michael Requa