For small medium repositories I suggest borg https://docs.openkm.com/kcenter/view/ok ... ackup.html
For large repositories might use different strategies it depends if you are in VM ( you can make incremential snapshopts with comercial software ), hight quality hardware provide automatic snapshots features and also there are commercial server - client backup software.
For large repositories might use different strategies it depends if you are in VM ( you can make incremential snapshopts with comercial software ), hight quality hardware provide automatic snapshots features and also there are commercial server - client backup software.