The true hot restore feature enables you to backup and store the hot snapshot of the server data on-premise with the help of Phoenix CloudCache.
In the event of failure with CloudCache, you can use the Backup to Cloud if unable to backup to Cache option on the Summary tab of the Configure CloudCache page, to back up the data directly to cloud.
- Select this option to back up the server data on Druva Cloud, if the Phoenix agent is unable to synchronize the backup data to the Phoenix CloudCache during a given backup time frame either because the Phoenix CloudCache server is not reachable, or it has run out of the disk space. This ensures that the server backups are always complete.
However, when you select this option, it is likely that all the backed-up server data during a hot snapshot might not be present on Phoenix CloudCache. Therefore, during restore, the data might have to be fetched from the Druva Cloud.
- Deselect this option if your organization prefers to store the hot snapshot data locally on Phoenix CloudCache. If the Phoenix CloudCache is unavailable during the backup time frame of a server, the backup fails after performing a certain number of retries to connect to the Phoenix CloudCache. For more information about disabling the option, see Configure CloudCache page.
Note: To enable this option, you must upgrade the Phoenix agents to version 4.3 or later and Phoenix CloudCache to version 3.1 or later. After upgrading Phoenix CloudCache, CloudCache backs up all deduplicated blocks in all subsequent backups.
After you upgrade Phoenix CloudCache to a version that supports true hot restore, you need to reattach the backup set to Phoenix CloudCache so that the latest backed up data is synchronized to the upgraded CloudCache. This will result in one time full backup where all data present in the backup set as well as Phoenix Cloud is synchronized to the Phoenix CloudCache.
Note: The data is resynced from the server to the Phoenix CloudCache. However, if the data from the CloudCache is never synced to the cloud, then the triggered decommission job will process in the next schedule.