Hello, I have read a lot of information on the forum about how to protect my data from HDD failure.
I am currently thinking seriously about creating a NAS to store my computer data every day or week to have a good backup for my work. I’m looking at using Seagate IronWolf 4TB drives to be future-proof. I have read that a Mirror uses 2 HDDs to store 4TB of data and RAID-Z1 uses 3 HDDs to store 8TB, with both solutions using one 4TB HDD to prevent failure. Is this true?
Also, I wasn’t able to find information about rebuilding a RAID-Z1 solution. How does it work, how do I do it, and how long will it take?
In answer to the “How long” part, it can take seconds for an empty pool. To hours for a fullish pool on medium sized disks, like 4TB. Even days for much larger disks, on full pools.
ZFS only re-syncs used data, so every disk replacement is unique, even with the same computer, disks and disk controller. It all depends on the amount of storage used, and it’s layout. Like an archive pool for video files might re-silver a replacement disk faster than an active pool for pretty small files, taking up the same amount of space.
Yes, that is true. You can go more technical than that but you don’t have to. It all depends on what your budget and risk tolerance is.
For example: If you think that 4TB in total is absolutely fine for you and you are sure you don’t need more in some years, go with your 2 disk mirror solution and a cheap(ish) 2-bay prebuilt NAS (and install TrueNAS on it) or build your own by putting 2 disks in a computer where TrueNAS can be installed.
If you’d like to be more flexible in the future and maybe learn a bit about zfs and RAIDZ and have the budget, go for a RAIDZ1 solution with 3 or more disks.
So what’s the difference then? Very simplified the difference is that a mirror is what it sounds like: 2 exact copies of your files on 2 drives. If one fails, the other still has all the data. RAID(Z) is a bit different. Here we talk about parity. This is a calculation of differences (again, very simplified) which is stored so that the data lost if a disk fails can be restored by using this parity data.
No matter what route you go my tips would be to buy a spare disk which you test and then keep safe and dry for the event that one of your disks fail. It is an investment but it is also a piece of mind (and judging by how crazy the market is right now, who knows if you’ll be able to buy a new one quickly if one fails). This is called a cold spare by the way. Also: If you can, build bigger than you think you need. Should you be needing more space in the future it is easier to already have the physical space for the disks and so on.
I’d argue (keeping in 4TB HDD size) that you can use 2 HDDs to store 3.9TB of data (or preferably less) as a mirror and RAID-Z1 with 3 HDDs to store 7.9TB of data (or preferably less).
Things get very bad when the pool is actually full; though they may (likely will) get degraded before they are filled.
With the “archive pool for video files”, the files likely don’t change much. Only metadata occasionally as new, large video files are added. Thus, less fragmentation and generally long runs of data blocks. Easier / faster to re-silver.
