Best practices for dataset layout on new installation?

[Chuck Remes]

Hey all, new FreeNAS user here. Currently building a beautiful box (X10SDV-TLN4F) now that all of the parts have arrived. Before I get to the point where I'm ready to install on that physical box, I thought I would experiment in vmware.

So, I created a (smaller) version of the box I'll be running complete with 7 hard drives setup in a RAIDZ2. I'm wondering if there are any best practices written up for how to layout the datasets on my volume ("tank," of course!). I did search the resources forum, how-tos but I did not notice anything relevant. If I missed it, please correct me gently. I've also been lurking on the forums for close to 6 months now.

Here are my intended uses.

1. NFS storage for my main desktop (OSX + Linux). My home dirs will be on each machine, but the NFS share will provide access to most of the data that I work with on a regular basis.

2. Plex server (in virtualbox, not jail). I have both movies and TV shows. I haven't added music yet but I plan to migrate that over soon. If possible I'd like to keep them separated for backup purposes.

3. Crashplan client (in virtualbox, not jail). Should have access to the data from #1 and #2).

4. 1+ FreeBSD and Linux virtual machines for development CI purposes.

My thoughts are that the layout would look something like this:

/tank/movies (all plex movies stored here and shared to plex vm via NFS)
/tank/tv-shows (all plex shows stored here and shared to plex vm via NFS)
/tank/music (all music stored here and shared to plex vm via NFS)
/tank/vm (storage for virtual machines, containers, etc)
/tank/home (my personal docs, development source, work in progress, etc)

Is this reasonable?

What should I look out for in terms of growth down the road... e.g. add business data which I would want stored on a separate dataset (maybe encrypted).

I'm trying to keep the folder/dataset hierarchy under "tank" relatively flat. Is that correct?

Any and all comments and suggestions are welcome. Thank you for your help.

 

[Robert Trevellyan]

The simple answer is, "Whatever works for you". The fact that you see a need to break things down, rather than just dumping everything in the top level dataset, indicates that you already have a sense of organization, so it really is mostly up to your personal preferences.

Beyond that, think about the fact that each dataset is like a self-contained filesystem. Thus many filesystem properties are per dataset, as are operations like taking snapshots and replicating them. These factors may help to inform your decision.

In my case, for example, all media are under one dataset, with sub-folders for tv and movies rather than child datasets. That's because I don't feel the need to treat them differently. You may have a different opinion.

Finally, it really isn't that hard to change things later, as long as you still have plenty of free storage.

 

[Chuck Remes]

Robert, thanks for the reply. I appreciate the detail about how every dataset is like its own filesystem. If I ever want to "zfs send" the filesystem to another location then having all of these items subdivided by kind will be very useful.

 

[agartha79]

See my post: https://forums.freenas.org/index.php?threads/dataset-layout-recommendations.57260