I have not thought this through yet but am curious to hear what others might think.
Two ideas, basically. The first is simply: how about using a Hetzner Storage box as Duplicati backend? Prices seem to be rather competitive, especially if you want more than 1TB storage. If that is an option, which of the many supported protocols would be the best to use?
But the really interesting thing (= idea #2) is this: Hetzner also offers some rather cheap VPSs and internal traffic between those and the storage boxes is free, which seems to make it possible to outsource some duplicati tasks (esp. checking hashes) to the server… Or to use Minio as your backend.
TBH, I have not yet understood the exact benefit of using Minio, but som competent people are promoting it as the ideal DIY solution together with Duplicati:
So, I might actually be bouncing three ideas here:
Hetzner Storage Box as backend, e.g. via sftp
Hetzner Storage Box with VPS running some kind of duplicati server
If you do not expect frequent large scale restores, Backblaze B2 seems cheaper up until 10 TB (at which point it comes down to USD EUR exchange rate…). Not entirely sure what the VPS would really buy you (unless you want to run some cronjob the regularly scans the whole repository which sounds interesting ). Looking at it it remains unclear to me how the data is actually stored and how much redundancy there is?
I am not entirely sure what minio would give you in this scenario…
Yes, I realize that the Hetzner prices are exclusive VAT, so if you have to add VAT, you are right. Plus, with B2 you are only paying for while with Hetzner you have to upgrade everytime your drive runs full.
All I know is that a number of automated snapshots of your storage are included. For example, in the case of the 2 TB box, you can have 8 snapshots (i.e. backups of your backup) per month.
Fine, as I said, I don’t really understand what I’m talking about.
That leaves us with:
Would that make any sense? Or would it fundamentally change the way duplicati works? The idea came to my mind because my impression from reading this forum it would be quite an improvement if duplicati didn’t have to download samples of the archive to check it but if the files would be checked server side…
Using a VPS to download and check the backups would obviously satisfy the ultimate in “are my backups actually there” paranoia (nefarious cloud provider might also just give you the hashes back that you expect if we ever get a feature that scans cloud hashes instead of downloading random files for spot checks) but in the mean time, I think I am better off using two separate providers (B2 and Onedrive in my case)
). Looking at it it remains unclear to me how the data is actually stored and how much redundancy there is?