I have been running Unraid on xfs pool for seven years. Unfortunately the hardware setup afforded by my Dell Poweredge R710 is starting to show its age. More and more current day software is requiring chipsets not present in the Xeon X series of processor requiring me to either work around or abandon applications. The single parity drive setup has saved me and my data numerous time, the ability to use various size of drives in an array is game changing. But now I must upgrade! Moving to the Xeon E series process solves most if not all of my chipset requirement. Also I was able to remedy another problem replaced to the cost of running servers, drive replace cost. The R710 is 2.5 inch form factor. The cost of 2.5 inch disk have skyrocketed. The newer unit shifts to a 3.5 inch form factor reducing disk drive replacement cost by more than half. Which get me to why I am posting here:

Under normal circumstances one would just pull the boot USB and all of the data drives, place them in the new unit, re-boot and be in business. But with the form factor change this isn't possible or (easily made possible). Since I am having to moving 10TB off the old array to setup a new array, it seems like a opportunity to migrate to zfs. Single parity xfs has been rock solid. Other than snapshots and replication Is this a lot of work for little reward?

Where are you replicating your snapshots to? Snap-shotting live data is risky and will not guaranteed to be functioning upon a rollback. Think databases. These need to be stopped and dumped to be safely restored.
What is your plan if you have more disks fail than a zpool can handle and you lose ALL the data? With the unraid array at most you only lose the data on the failed disks and not the entire array.

Do you want ALL your disks spun up and reading/writing 24/7? There is no selective spin down.

Do you need more than 200+MB/s read speed on any single given disk?

How familiar with zfs filesystem are you?

Do you have excess RAM you're needing to put to use? ZFS performs optimally the more RAM you can allocate to it for read caching. It's not a hard requirement but it definitely help to have more RAM for zfs.

IMO there is a time and place like using zfs on ssd pools but for HDDs unless you know you need the speed and throughput (enterprise workloads) single disks in XFS is my preferred way to go.

Great questions,

1) I have experience with zsf in a Truenas environment (my cli experience is limited)

2) zpool data lost of more than one disk would be disastrous

3) system would have 48gb of memory, my truenas box has 32gb and doesn't seem to bottleneck

4) don't want spun up disk 24/7

xfs has been rock solid, sounds like your saying "if it ain't broke"!

5 hours ago, holtdw01 said:

XFS has been rock solid, sounds like your saying "if it ain't broke"!

Pretty much yeah