Unless you configure otherwise, the default location for pretty much all the important stuff is /opt/zimbra. While there's other things modified (system users/crontab/initd/syslog/logrotate/prelinks/rpmconf/etc) and some logs are stored elsewhere like /var/log, /opt contains all your data openldap, mysql metadata, blob store, lucene index, config, etc such that if you had / & /opt partitions you could replace the OS on / (configure it practically the same, matching hostname, UID's, etc) and run
install.sh -s to get a working system.
If you want a break down how much to assign to / /boot /swap /tmp /var /opt /usr we can provide it, but I'm just playing with the bare min above. It's possible to do things like put queues & spools on ram disks, mysql & ldap on a fast disk, indices on a different partition, and stores in another location, redologs wherever (and if NE use HSM to move old items to even slower disks or put the hot backups elsewhere). Use mount, symlinks, or
zmvolume to your hearts content.
One Zimbra install per OS, and I gather by the above convo these are physically separated not virtual machines, nor are you linking them in a multi-server or multi-domain setup, just individual self contained instances.
I would hope that 160GB is across 2x80GB RAID1, otherwise your point of failure is pretty high.
If it's 1x160GB I might even do lvm mirror or mdadm 2 partitions to help with unreadable bad sectors (can't remap if you can't read it) but that cuts your space in half. Best to have some type of frequent
backup solution in place.
Course if it's really 1box+hostOS X 6vm guestOS's X 160gb allotted each on some underlying RAID10, and you just wanted to make it simple in explaining, then cool.
Bugzilla
Wiki
Source
starter questions 
Linear Mode
