


Besides Proxmox Backup Server, every datastore also accepts tools that work over SSH: SFTP, rsync, Borg and ZFS replication. They all log in with an SSH key that you add to the datastore once.
Open your datastore, click Connect and select rsync / borg / sftp:

Paste your public SSH key and click Save. The dialog also shows the user (customer<ID>) and the host that your tools connect to. Test the login before you set up a tool:
ssh -v customer<ID>@<host> -i /path/to/private-key
The examples use customer<ID> and <host> as placeholders. Replace them with the values from the Connect dialog.
Borg stores deduplicated, encrypted backups in a repository on your datastore. Create the repository once, then create an archive for each backup:
# Create the repository
borg init --encryption=repokey customer<ID>@<host>:./borg
# Create a backup
borg create customer<ID>@<host>:./borg::{now} /path/to/files
List your archives with borg list, and free space after you deleted archives with borg compact:
borg list customer<ID>@<host>:./borg
borg compact customer<ID>@<host>:./borg
rsync copies files and folders, and on later runs only what changed. Pass your private key with -e:
rsync -avz -e "ssh -i /path/to/private-key" /home/user/documents/ customer<ID>@<host>:/backups/documents/
-a keeps permissions, owners and timestamps, -v shows what is copied, and -z compresses the data during the transfer. rsync makes a copy, not a versioned backup: a file that you delete or damage locally is gone on the datastore after the next run too, unless you use --backup or a tool like Borg.
Any SFTP client can upload and download files. Use your user, the host and your private key:
sftp -i /path/to/private-key customer<ID>@<host>
Graphical clients such as FileZilla or WinSCP work the same way: choose SFTP as the protocol and load your private key in the connection settings.
You can replicate ZFS snapshots into your datastore with zfs send and zfs receive. Received datasets must go below zfs/customer-<ID>/recv:
zfs send localpool/dataset@snapshot | ssh -i /path/to/private-key customer<ID>@<host> "zfs receive zfs/customer-<ID>/recv/dataset"
Send an incremental stream with zfs send -i on later runs. zfs list, zfs get, zfs create and zfs send also work, but only inside zfs/customer-<ID>/recv. ZFS is available to the main login only, not to SSH users.