My current setup is an OMV 5 installation with 4 drives aimed at data that can be lost without too much pain (going to be more, I'm still in the process of setting things up). They are XFS formatted and joined together using MergerFS. SnapRAID may be added at some point, for at least a minimum of redundancy.
Then there's three 8 TB drives that I want to use for data that need more protection. This will grow over time. One is a SMR drive that I want to use mostly for home videos - large files, that do not change over time. The other drive is PMR and to be used for photos. I might end up using this drive to work directly on, but it's also possible that I will move files that I want to work on to a local SSD for performance reasons, and after that move them back to the NAS. Possibly a mix of both uses, depending on what I want to do. The third drive is to provide the redundancy.
I would like to have versioning, but without block level deduplication it's not practical enough.
What I am thinking so far: ZFS/BTRFS as one volume sound enticing, however I do not have control over which drive is data stored on, rather it is spread out over all drives. If I lose too many drives, all is gone. That sounds like a terrible idea.
But what I could do (I guess) is use BTRFS or ZFS for each drive on its own, and then use MergerFS and SnapRAID on top of that? And set up the file system to do the snapshots...? That way I can undo errors, worst case scenario, recover from nasty ransomware...? Would this even work? I suppose there are no easy to use management tools in OMV? I guess I could do something similar with a more regular file system and LVM, but it would waste a ton of space. How would BTRFS behave when I remove data and add it back (in slightly modified form, i.e. some metadata in the file is changed) later? I suppose I need to use one of the deduplication tools available for BTRFS, but do they also work with data that's inside a snapshot?