So I’m working on revising my NAS, and I’m trying to find a way to make the actual access to the NAS faster and still have resiliency of hard drives.
Right now, I’m using ZFS on Ubuntu with 3 disks in a RAIDz1 (2 for storage, 1 for parity). Each drive is 8TB so I have a total storage size of 16TB after the one drive being used for parity.
The issue is that throughput is really low - as expected, since these are 3.5" WD Red’s which although made for NAS use are definitely not fast.
I figure I can build a different pool of just SSD’s, but I want to keep using my already existing HDD’s.
I figure I can have a setup like this:
- Keep the already existing 16TB pool (let’s call this “old pool”)
- Add enough SATA SSD’s into the system to reach roughly 16TB of storage (let’s call this “new pool”)
- Use the new pool for actual read/write operations
- When the new pool is not actively in use, push all he changes to the old pool
From trying to read over the different ZFS wikis and such, I could just create a rather large SSD set for ZIL / SLOG and a separate SSD set for L2ARC - but I don’t believe this will do exactly like I’d want to.
Is this possible to do with just plain ZFS on Ubuntu? Or would I have to move to another OS like TrueNAS or unRAID?
Also I feel it’s important to mention that I try follow the 3-2-1 rule by having this data stored on the NAS as well as the cloud, but I’d still be missing one local place to actually follow that rule - would this idea count or is it considered a a single location?