I have a media center(Separate from NAS) and large size video files (50 to 100 GB each) that I want to store on NAS.
Its for home use and I am the only user so Performance is not priority. I do have the video files on backup Disks but I would rather not spend hours/days retrieving files from backup in case pool goes down. I don’t want to lose complete data.
I want to create a single network folder visible to media center where all the files are accessible.
At present I have four 4TB HDD and plan on adding more disks and possibly increase size in future as price goes down.
After reading and watching videos on ZFS Storage what I understand is
Say a vdev goes down, then the whole pool is lost, so having multiple pools is the only option?
Also you can only add/remove vdev if using mirror?
Is there a way to create multiple pools and have it appear as single network folder to someone accessing these files. I need to give this path in Media Center and I want to use single path (single folder).
You build your vdevs with redundancy, e.g. as a mirror or a RAIDZ2. In a RAIDZ2 vdev two drives can fail and the vdev and your data will still be present. Of course you should replace the first failed drive as soon as possible to restore the desired level of redundancy.
I don’t understand your remark about multiple pools. Only option for what exactly?
A pool can consist of many vdevs, each vdev providing the (recommended) same level of redundancy. E.g. 4 vdevs of 6 drives each in a RAIDZ2 configuration. 2 disks from each vdev may fail without data loss.
All your other requirements can be met with a single RAIDZ2 vdev or RAIDZ3 if you want it extra safe.
“You build your vdevs with redundancy.”
“A pool can consist of many vdevs”
Yes I am aware.
I am worried about
Losing more disks during resilver due to bad sectors. I am using used disks and doubt the reliability.
Accidently messing up vdev/pool with some stupid mistake as I am still learning. I would love to save some data, so I was thinking of creating multiple pools rather than having single pool. In case one vdev/pool goes down I would still have some usable data left.
I want the option to remove vdev which RAIDZ1,2,3 wont give me. Please correct me if I am wrong.
I don’t understand your remark about multiple pools. Only option for what exactly?
I do not want to lose complete data. Partial data loss is acceptable but not complete. So I was thinking of creating multiple pools rather than copying all data into one pool.
Example I have 6 disks and I create single pool with three vdev with mirror, if I lose one vdev I lose all data. But if I create three pools with one vdev per pool in mirror and if I lose one vdev/pool. I would still be left with two usable pools and would need to get backup from dvd only for the data on lost pool. But in creating three pools I believe I am creating three different network folders?
How can I show them as one folder to users, if this is possible.
I might be thinking wrong, intent is setup such that I do not lose complete data but maybe lose partial data.
Strictly speaking, RAID(z) is not a backup(z) backup. It is a tool for high availability. With the higher parity drives number having the higher availability.
If you care about your data, you should consider implementing a 3-2-1 backup strategy. It won’t be cheap for the video files, though.
“Strictly speaking, RAID(z) is not a backup(z) backup.”
Yes I am aware and I have backup on DVD.
“It is a tool for high availability.”
Exactly my point, if a vdev goes down whole pool goes down. Nothing is available.
If I used 4 disks directly in linux or windows, if one disk goes down I would still have access to other three disks. In ZFS if a disk goes down I can recover but If one vdev goes down, I lose complete pool and needing to spend hours/days to get data from backup. That is why I mentioned creating multiple pools.
Pooled storage is one of the key features of ZFS, so by creating multiple pools you’re giving that up. You’re taking on a management/administrative burden that you don’t need to.
Availability is about uptime, not about data-loss protection. You cannot bring it to 100%. Period. Even the best of the best with almost unlimited resources can’t. You will lose your (1st copy of) data eventually. Even if we do not take ransomware and (un)natural disasters into account.
By splitting the data into multiple pools, you would bring average uptime a bit higher. Well, it depends on how we calculate it. However, you would lose:
Ease of management as pointed out by @dan. Including storage management and shares management.
Overall performance. Your storage performance would be ~N times worse. With N being the number of pools.