So the script is tested and supported on a CentOS 6 host so you can run it on any CentOS 6 host. The reason we recommend another host is simply so that you can have the backup files on another device should something fail on the upgrade on the Admin Server, for instance. If you try to reimage a physical host for a second time, you may lose that preserved partition that we have in the case of physical upgrades.
The idea is simply that you have backups of your backups. If you take the backups you create on the Admin Server and store them somewhere else in case of the worst, it will be fine.
And just like John Snider mentioned in your other question,should the worse happen or you are not attaching the drive that has the backup on it, you can SCP the backup over if you go ahead and configure the networking ahead of time.
Absolutely! You'll need some space to store the backup information on the head unit, but that shouldn't be a problem unless you have an abnormally large amount of data on that appliance. If you run the script with the "-d" switch, it will check every appliance for space before the scripts run. Using the "-D" (capital D) will run a more comprehensive disk check, but only bother with that if you've run with the "-d" test and it fails.
I recommend that after you've copied the scripts to the head unit (create the /var/netwitness/database/nw-backup/ folder and put the scripts there), run the get-all-systems.sh script, then run ssh-propagage.sh script, and finally run nw-backup.sh by itself without any switches. That will give you all of the options you can use.