Our main storage pool consists of four 3TB disks, so we have 12TB raw disk capacity. For redundancy we use ZFS’s zraid (which is basicly a RAID5) as layout. This puts the usable space of the disk pool in the area of 9TB (4 x 3TB – 3TB).
Since we probably have lots of duplicate and compressible data in the backup and archive datasets we enabled deduplication and compression in this two and a few other datasets. Now after a few months we reached a funny point with the growth of the pool.
The volume of all the things we put in the pool (when we undo the compression and dedup in calculation) now exceeds the raw space of four 3TB disks and we have still 35% free space. Nevertheless we have the redundancy that one disk can fail without data-lost.
The downside of dedup & compression is, that it can lead (and have led) to serious performance issues with a heavily used pool. So far we solved this issues by fine tuning ZFS and schedule big dedup-write jobs during the night.