I’m looking for suggestions on how to configure my backups. I have an empty local drive and a OneDrive plan with enough storage.
My idea was to backup to both the local drive and to Onedrive, one being fast but “unsafe” and the other being slooow but for sure more “reliable”.
Which one would you suggest, backup with Duplicati on the local drive and then carbon-copy the local files to the cloud (with rsync or custom Python script), or create two identical backup jobs but with different target locations?
I have the idea that the first one is somehow faster, as it’s a brainless copy of the backup set. But at the same time any error is propagated, such as if local files get corrupted without notice, the automatic script would also corrupt the cloud files.
The second option would require Duplicati to do twice the same job.
Do the same job twice with different remote storage targets. While Duplicati supports verification of backup to detect corruption, depending on the size of your backups this could take a long time. By default I think it checks a single dblock file with each run.
I would do a sync (eg syncthing) to the local drive and backup to the remote drive. Duplicati is designed for untrusted storage. I assume you trust the local drive, so I would not use duplicati for that. Unless you plan to store the drive at relatives or friends.
Benefit is you have easy and fast restore (nothing needs to be decrypted and put back together).
Both @samw and @Wim_Jansen have valid suggestions - it really depends on your needs, hardware, and bandwidth.
A local sync of the raw files as @Wim_Jansen suggests definitely has the benefit of a fast restore, but that restore has to come before your sync tool (such as syncthing) has syncronized the file change or deletion from you main drive to your synced drive. Additionally, unlike using Duplicati to the local drive, a sync tool won’t have any version history locally backed up.
If you don’t need version history, then the local sync may be the way to go (though I have to wonder if a mirror RAID might work just as well).
If you want version history, then a double backup as @samw suggests would probably be the way to go where the local Duplicati backup would be your first (and fastest) restore point.
Other options include:
2 jobs, one to local to USB and one remote. This allows for removal of the USB dive when not backing up providing an air gap against infections as well as potential electrical hardware issues. Additionally, different schedules can be done (for example daily to remote and “when I think of it” to local)
1 job to local and a sync (such as rclone or syncthing) of that backup to remote. This gives both a local backup (faster restore) and offsite storage security with the drawback of if you have to restore from the remote local you have to somehow get the files copied locally OR point Duplicati to the remote location
I’m not a OneDrive expert, but I expect you can’t run rclone or syncthing to OneDrive, but could you have your one drive contents be a folder on your local machine? If so, then having Duplicati back up to your local OneDrive folder would ALSO store the files on OneDrive (a drawback being that if they are deleted from OneDrive, they get deleted locally - unless OneDrive allows control of directional creates/deletes).
Hopefully that all gives you some stuff to think about!
@Wim_Jansen the option to sync to the local HD is not viable, as I have a ciphered hard drive on my laptop and don’t want to have a clear copy of it around the house. I can cipher the whole HD with things like BitLocker, but then I’m not sure if I can attach it to the USB of my router and access it via Samba or similars (that was my original idea). I had a Western Digital self-encrypting hard drive but no way that the Vodafone router (bleah) I have at home could handle drive encryption.
@JonMikelV the option to have a OneDrive synced folder in my PC is very clever! I think I can also create that folder in my external HD if I manage to attach it with Samba, which would solve the drive size issue. I’m not sure about the control of directional creates/deletes as you mentioned, but OneDrive has versioning and stored copies of deleted files, which would make it easy to recover them.
samw really? I thought tha Duplicati had some sort of control in place, like checking hashes of backup files or similars If that’s the case, then 2 jobs are for sure the best solution
@dgasparri it does. But by default it only checks a single dblock file. You can change it to check it all, but that would require sucking down your entire backup from the destination.
If that’s the case, then 2 jobs are for sure the best solution