Replication between two datacenters
- Thread starter akjome
- Start date
With ZFS this would be possible with pve-zsync https://pve.proxmox.com/wiki/PVE-zsync
Another or additional method might be to have one ProxmoxBackupServer on each site and sync their backups between them. So in case of a disaster, you would live-restore the Backup on the second location: Thus the VM would start and already run while the restore is continued.
Thanks @Johannes S. I am going to test the PBS option on each site. Need to find out how to get the live restore option working
Live restore is just built-in, presuming you have a fast enough storage and network, it is relatively invisible. Note that you have to make sure the PBS remains available during the restore, otherwise if it fails, any writes between the restore and the failure will be lost - which is not a big problem for static servers (eg. If you don’t store home directories or databases on disk). It is a risk you may want to re-consider for eg. huge database systems that may take hours to restore.
It's not the same as real replication though since in case of a disaster you would still first need to start the restore. While pve-zsync actually syncs the data and configuration of your VMs between the clusters and usually more often than the regular backup schedule (default is 15 minutes for pve-zsync while most people tend to do only a daily backup). The consequence is that in case of a failure of the first cluster, you should be able to launch all failed vms on the second cluster with minimal data loss/downtime.
So my approach would be to actually do both: pve-zsync to have a up-to-date replication of the vms and PBS so I can only go to an older point in time (in case of a hacker attack or failure of the pve-zsync job).
Last edited: