Restoring Full Instances for Google Cloud Platform

Updated

You can restore one or more instances at a time. Instances can be restored in place (overwriting the source instances) or out of place with a different instance name, or to a different Google Cloud Platform (GCP) service account.

Full instances can be restored from a streaming, IntelliSnap, or backup copy.

Notes

  • The Private IP address of the source instance is included in full backups. For in-place restores, if that Private IP address is available is restored to the instance. If that Private IP address is not available, then a new Private IP address is automatically assigned to the instance. To restore Private IP address information, you must add the compute.addresses.useinternal permission to your GCP service account. For more information, see Creating a Google Cloud Platform Service Account.

  • For in-place and out-of-place restores, you can use shared virtual private cloud (VPC) networks.

  • For out-of-place restores, if the instance selected for restore includes regional disks, all of the regional disks are restored to the selected zone.

  • If you want to restore a GCP instance that was turned off during a backup operation, you must assign that instance an external IP address using the GCP portal.

  • From streaming backups or backup copies of encrypted disks, you can restore full instances, files or folders to the same project or to a different project. During the restore operation, the Commvault software verifies whether the source encryption key ring and key are present in the project that you are restoring to. If the key ring and key are present, Commvault assigns them to the new instance disk. If the key ring and key are not present, Commvault generates a new key ring and key (with the same name as the source encryption key ring and key) and assigns them to the new instance disk.

  • If you want to restore an instance without changing the project or zone of the instance, if the backup VM is in the sole-tenant node, then the restore operation is performed in the sole-tenant node.

  • For out-of-place restore operations to different project or zones, when the source instance is part of a sole-tenant node, the restore operation operates without the sole-tenant properties.

Before You Begin

  • Verify that your GCP service account has the appropriate roles and permissions to perform the operation. For more information, see Creating a Google Cloud Platform Service Account.

  • To restore an instance to a different GCP project, the service account of the virtualization client must have access to the different project.

  • When restoring out of place, you must specify the destination zone, machine type, and network settings.

  • If a passkey is configured for restores, you must have the passkey.

Procedure

  1. From the navigation pane, go to Protect > Virtualization.

    The Virtual machines page appears.

  2. On the Hypervisors tab, click the hypervisor.

    The hypervisor page appears.

  3. On the VM groups tab, in the row for the VM group, click the action button , and then click Restore.

    The Select restore type page appears.

  4. To restore from a specific copy of backup data, in the upper-right corner of the page, click Settings, and then from the Source list, select the copy.

    If you leave select Automatic selected, the restore operation searches for the requested data in the primary copy, and automatically selects a different copy if the data is not found in the primary copy.

  5. Select Full instance to restore a complete instance to GCP.

    The Restore page appears.

  6. Select the objects to be restored. Select an item or click on an entry in the Name column to browse within an item.

  7. Click Restore.

    The Restore options dialog box appears.

  8. Provide the requested information:

    • Type: Select one of the following options:

      • In place: Restore to the original location.

      • Out of place: Restore to a different location (with a different instance name, or to a different GCP service account).

    • Destination: To restore to a different GCP deployment, select the hypervisor from the list.

  9. Under Access node, select an access node for the restore, or select Automatic.

    When you restore multiple instances or VMs, select Automatic to distribute the workload (instances or VMs) in the restore job across the access nodes (also called proxies) that are assigned for the destination hypervisor.

    First, the instances or VMs are assigned to the access nodes by a zone match. Then, if access nodes are not available in the same zone, the instances or VMs are assigned by a region match. If access nodes are not available in the same region, then the instances or VMs are assigned using the round-robin method.

    The Automatic option to distribute the workload is available for full restore and conversion operations, both from streaming backups and IntelliSnap backup copies.

    • When restoring out of place, you must specify the following:

      • Instance display name: (when restoring a single instance) Type a new instance name to be used for the restored instance.

      • All VMs: (when restoring multiple instances) Select a specific instance from the list so that you can type a new instance name to be used for the restored instance.

      • Zone (also called Primary Zone): Click Browse to select the destination zone for the restored instance.

      • Secondary Zone: Select a secondary zone for the restored instance.

        This option is applicable only if the source instance has a regional disk configured.

      • Machine type: Select the machine type for the out-of-place restore.

      • External IP: By default, this option is enabled. Clear the check box to disable.

      • Network Interface: Select the network settings.

    • Power on VMs after restore: Select this option to start restored VMs automatically.

    • Unconditionally overwrite if it already exists: To delete an existing instance and replace it with the restored instance, select this option.

      Note: If an existing instance with the same name exists on the destination host and you do not select Unconditionally overwrite if it already exists, the restore job fails.

  10. Click Submit to run the restore job.

For information about sole-tenant nodes, go to the Goggle Cloud documentation website, and search for "sole-tenant nodes".