this post was submitted on 24 Feb 2025
4 points (100.0% liked)

unRAID

1197 readers
1 users here now

A community for unRAID users to discuss their projects.

founded 2 years ago
MODERATORS
 

This is a follow-up of my previous post where i observed that enumerating the files on a unraid share with a million files took 10x the time it took vs the same stuff shared via truenas.

I now removed all the drives in the array (new feature of unraid 7) and exclusively have a btrfs "cache" of the same size. It is less "efficient": unraid xfs array with 8x 4tb drives gave 24tb usable space with 2 drives redundancy. Now with btrfs i have 16tb usable space with only 1 drive redundancy. But the SPEED difference is insane.

Before, syncing a 1tb directory using unison, and the contents were already present on both servers: 3 hours

Now: 30 seconds

top 1 comments
sorted by: hot top controversial new old
[–] [email protected] 1 points 5 days ago

No real surprises here

unRAID's MD device array provides very high flexibility in terms of storage options (being able to use any drive size in the array that is smaller than parity), and preserving individual array drives in the case of a failure that exceeds the capability for parity protection, rather than losing all data on the entire array (you'd only lose data on the drive that failed).

The price of all that flexibility is speed.

For many types of array data, that's not a big issue, but occasioanlly for huge numbers of files, it's very slow (especially for some lighter-weight systems).

Using a ZFS striped array in unRAID (same as TrueNAS) should provide the speed you're looking for, but you lose that amazing ability to just add drives and expand/change your storage needs almost effortlessly.

The great thing is that you can have your cake and eat it too!

My array has 13 unRAID-style array drives with dual parity, protecting all of my media (who cares if it's a bit slow, just TV shows, Movies and Music).

All of my important stuff (bank documents, kids stuff, family photos, etc) is on a much smaller ZFS array that's replicated hourly with snapshots to a second independant unRAID server, and from there, weekly to a tiny 2-drive ZFS array that gets rotated-out to a safe deposit box every few months. I used to use rsync to do that, which was slow and painful. With ZFS and snapshots, it's effortless and super-quick.