I am redoing my backup architecture and am considering ZFS replication. I presume the snapshots are what provide the versioning capabilities in a setup like that and I suppose additional snapshots can be taken on the target machine to increase the number of versions.

However, this workflow for a backup seems a bit foreign to me. Maybe it just seems too "simple" or "easy" for me to think it'll work. Currently I use Borg as a backup solution and I'm pretty happy with it. Perhaps this is just the windows admin side of me saying that if I don't have to fight with it, then it ain't working.

I'm looking for thoughts on relying on ZFS replication and snapshots as a backup. Or should I stick with something like borg?

I wouldn't consider replication and snapshots a true backup solution, but I guess it also depends on what you are looking to protect against. For example replication is great if there is a problem with a single server (or site), then you have a full up to date copy of all your data on a different system. As a backup method though it does nothing if you accidentally delete some files as they will get deleted on the replica system as well.

Snapshots do offer some more protection though, and fast recovery of small subsets of the file system. How viable this is for recovery really will depend on a number of factors. I will generally look at snapshots as a take a few through the day, keep maybe a week worth, and it is a simple fix to the accidentally deleted or altered file. However if the problem was a corrupted file system it is possible that snapshots won't help.

For anything important I would always recommend doing a full true backup to another system (or ideally multiple systems so you have a copy on a different media type and a different location). At the end of the day it will really mostly depend on how important the data is to you, and how much can be invested in protecting it.

2 members found this post helpful.

+1 to all of this. Replication only helps so much...what if a file got corrupted and you need a copy from a week ago? All you have is the same corrupted file in two places, but a genuine backup gives you a broader safety net.

I always replicate for servers that need to be up 24/7/365, but ALSO have full and incremental backups on other media. I typically take a full backup weekly, with nightly incrementals.

1 members found this post helpful.

All of that totally resonates with me. I suppose there may be an argument to be made for the ZFS resiliency and snapshots with regards to file corruption at both a block level and at file level respectively. There may also be an argument to be made for the simplicity of leveraging existing filesystem capabilities for the backup solution. Though this is a home lab and I'm tempted to experiment with using ZFS in this way, I don't think I'm brave enough to rely on it over a more traditional backup job.