Backup, Restore and Instant Restore Working Principles with OpenStack

Working Topology

Principles of Backup

1.When a backup job started, Vinchin Backup & Recovery (hereinafter referred as the backup server) will request hosts information from the OpenStack cloud by using OpenStack API (token-based authentication by Keystone is required for API requests), the process will be done over the management network. And then communicate with the backup agent installed on the hosts, to obtain detailed XML config files. By combining the above information, the backup server will get to know the location and type of the root disks, ephemeral disks, and cinder volumes of the VMs.

2.According to root disk and ephemeral disk type, request disk snapshot by using storage API, as for cinder volumes, cinder API will be used to create snapshots for the cinder volumes.

3.After snapshot, the backup will start. If the root disks and the ephemeral disks are on the Ceph storage cluster, the Ceph config file will be downloaded to the backup server by using the backup agent. Then the backup server will parse the downloaded config file and use the storage pub network to access the root disk and ephemeral RBD snapshots in read-only mode by using Ceph API, when done, the backup is completed.

4.If the cinder volumes are centralized LUNs, the backup server reports its WWN to the storage device over the storage network and requests to mount LUN snapshots to the backup server in read-only mode. Then the backup server will read the snapshot and backup. Once completed, the backup server will send a command to the storage network to unmount.

5.After backup of the root disks, ephemeral disks, and cinder volumes, snapshots will be deleted according to the backup mode.

Principles of Instant Restore

1.Backup server shares the backup repository to OpenStack by NFS as the backend cinder storage.

2.By using OpenStack API to create cinder volumes for the VMs to be created and then create VM.

3.Start up the VMs to bring up the production environment, at this moment all VM data is still on the backup repository.

4.By using the migration feature of the instant restored VM to migrate all data to the OpenStack production storage.

5.To stop the instant restored VM, the VM and the cinder volumes will be deleted automatically, if there's no other instant restored VMs then the NFS shared storage will be removed from the OpenStack cloud as well.