Roses are red ๐ŸŒน
Please try to keep up
Wtf is all this talk in my timelines
About RAID being a backup?
@jerry forgot the truck on fire, Mr. Llama

@jerry

It's not DNS.
There's no way it's DNS.
It was DNS.
-- a DNS Haiku

@jerry a couple of days ago I stumbled upon a discussion about restic where somebody said they didn't need a dedicated backup tool because with ZFS they could just use snapshots
@jerry oh is it the usual suspects, again demonstrating they are unqualified?

@jerry raid is not backup, it is (or at least in certain configurations can be) resiliency.

Who took crazypills and thinks otherwise?

@fennix Iโ€™m not sure. I just jumped on the angry wagon because it seemed like the right thing to do
@jerry it's possible, keep raid in always degraded state attach disk rebuild/mirror daily? and detach disk ;) rsync via raid....
@nebula1000 yeah, local offline backups can be good a thing, but this would presume the drives donโ€™t burn up or get flooded out or otherwise die.
@jerry So... what's a genuine backup then? I got a RAID 1 thing going, with the most important data mirrored/synchronized on three devices. I hoped it would suffice, but am eager to learn what I could do to make it safer.

@Gilgamesch @jerry RAID is fault-tolerance. It allows a drive to fail without disrupting your data.

What if you delete a file and you need to get it back? The deletion is replicated immediately to both drives in the RAID set. A copy-on-write filesystem like ZFS or btrfs can help here, because you can take snapshots, which lets you undo file changes like deletion with a certain granularity.

What if thereโ€™s a fire while youโ€™re out and all your drives get turned to slag? This is where you need entirely separate backups not connected to the main system. A backup drive or NAS at the same location is better than nothing, but a fire could catch those too. An off-site backup is a bigger hassle, but itโ€™s much less likely to fall to the same incident which took out the main copy of the data.

@Gilgamesch to be very clear, there is a benefit to raid. Especially with spinning disks, hardware failure is one of the big causes of data loss and raid will help mitigate that. But things like ransomware, fires, accidental deletion, and so on arenโ€™t helped by raid.

I would pair raid with an off-site backup solution using snapshots, with a tool like borg that makes an offsite, encrypted backup that you can restore from in more catastrophic circumstances

@jerry

@Gilgamesch

unwritable snapshots help with ransomware.
befor you can ask what is a backup you need to define the threat. if you have backups in the datacenter and the center burns with those backups you don't have backup. I can come up with scenereos where all the locations you have backups are destroyed at once.

Raid wich snapshots will cover the majority of your backup needs. It won't cover all but if done well it gets you into that is stastically unlikely for many so maybe you can stop worriying there??? Of course not all data is worth saving in the first place while other is priceless whichiagain points to different backup.

because of the above I conclude anyone saying raid is not a backup doesn't uneerstand the real problem.

@Gilgamesch @jerry
1. RAID as a "hardware failure backup"
2. Synchronization (with versioning and a recycle bin) as a "first tier backup"
3. Offline backup on an external hard drive as a "second tier backup" (better: off-site too) - first ransomware protection, protection against fire, water, ...
4. Append-only Cloud backup with restic (encrypted B2 storage with delete protection and data retention) as "third tier backup" (full ransomware protection due to the nature of the backup config)
@jerry, here is mine:
RAID is no backup,
No more than your NFS.
Testing is caring.