For me personally, my 5TB x 2 RAID1 (ZFS really, so mirrored) configuration with 5TB usable is enough for my personal uses. If I need more space, I'll configure my ZFS with +2 HDDs for another pair worth of space+redundancy.
For me personally, my 5TB x 2 RAID1 (ZFS really, so mirrored) configuration with 5TB usable is enough for my personal uses. If I need more space, I'll configure my ZFS with +2 HDDs for another pair worth of space+redundancy.
The reason being that for larger drives, resilvering can take so long that secondary failures become probable or—in bad cases—even virtually guaranteed (This is the same probably the WD SMR drives had, but it exists even on normal drives if they get big enough).
For this reason, for modern large drives, most people recommend at least RAID6 (known as RAID-Z2 on ZFS), which means you have two drives worth of parity as compared to RAID5's single one.
Of course, this means that you need at least four drives for it to be worthwhile, and most people realistically end up going for around six disks, otherwise you are spending a lot of your total cost on parity.
For my home bulk storage, I just have two normal 14TB drives and a nightly cron to rsync archive (new files only) from the primary to the backup. This gives me enough of a backup for both user error and hardware failure. It doesn't protect against a house-fire but that is ok.
I should maybe have considered BackBlaze but I can't believe they are going to last long if they continue offer flat-rates for multi-TB attached NAS.
Mirroring is for preventing hardware failures.
ZFS Snapshots are one protection against user-error. Backups are another.
It may be simpler to play with ZFS on VMs. The VM-disks will be small and inefficient, but "good enough" to prove whether or not the system works.
Not that I've tried any of this before. I happened to have an old computer lying around with ~4 small hard drives that I messed with to learn the ins-and-outs of ZFS the first time I used it. (Shuck hard drives, pull them out of laptops, etc. etc. Its not really that hard to get 4 hard drives to play with, especially if you ask your friends for ancient laptops with 200gb drives or something)
Of the many ways storage can get complex, I've found ZFS to be one of the most straightforward.
They're not exactly cheap, but I'm another one who went the Synology route. They slap a user-friendly UI ontop of linux mdraid and BTRFS, so you can RAID together a few drives, set up automatic BTRFS snapshots with retention rules (keep 12 hourly snapshots, 7 daily, 3 monthly, etc), set up disk scrubbing schedules etc all just by clicking around.
So it seems dizzying and unstable, but the dizzying bit is just nerd ego and hype, and the instability is also just nerd ego meets inevitability. (Now I think about it, it's a bit reminiscent of the abandoned "ABC shall be the ultimate XYZ" software you sometimes find online, and scriptkiddie mentality.)
ZFS is just a reasonable, mostly solid, somewhat idiosyncratic filesystem with nice features. Most of the failure noise online is from people who have no idea what they're doing; there are curiously (good-suspiciously) few public reports of commercial-scale data loss et al.
I've taken to using it as the root filesystem on a couple of Debian boxes. The Debian install instructions are both overblown and also almost hilariously broken, but that's a devops/sysadmin documentation failure, and not ZFS' fault. I spent a few years on Slackware, and have found the ZFS setup experience reminds me of that rock-solid-meets-flying-by-the-seat-of-ones-pants mentality.