While the exact process may vary from database to database, for a typical database operation, a full or incremental backup is taken nightly, with regular backups of the archive logs about every four hours throughout the day. The recovery process requires restoring the last full backup followed by the required incremental backups. You then apply the incremental backups and archived logs to recover the database back to the current point-in-time (PiT). This is a time-consuming process and can extend the period that the database is unavailable.
By doing regular database snapshots using PowerFlex Storage Snapshot functionality, you can facilitate fast recovery from certain database failure or corruption scenarios. The recovery time depends on the regularity of the snapshots and the database change rate. However, a lengthy restore from a remote backup location is not required because the snapshots can be restored instantaneously. You only need to apply the archive logs and online redo logs generated since the last snapshot. This results in a simplified and rapid recovery process with minimal steps required.
Simultaneously taking snapshots of all contributing PowerFlex volumes automatically creates a Snapshot Consistency Group.
In this example, the volumes are SNAPDEMO_DATA1, SNAPDEMO_DATA2, SNAPDEMO_FRA1, and SNAPDEMO_REDO1.
You can use the default snapshot name and suffix or create a custom name format. For snapshots that are only required for recovery purposes, we recommend that you ensure the Read Only option is selected (default).
The following dialog displays, and the snapshots are created with the names and suffixes specified.
The system instantaneously creates four new snapshots volumes. This is the snapshot set that you will use for doing recovery in this use case.
The table DEMO_TAB now has 20,000,000 committed rows.
You can simulate a corruption or media fault in the database by deleting one of the databases data files. Subsequently, a database restart will show an error as shown in the following figure:
In the current use case, the data files are stored in the SNAPDEMO_DATA disk group.
Note: While all the data disk groups and volumes can be restored from their respective snapshots, this requires more media recovery and takes longer. Our recommendation is that you restore only the snapshots that are required. Do not restore the REDO and FRA disk groups.
The following verification dialog is displayed:
A final verification dialog is displayed.
This confirms that the snapshot (taken at 10,000,000 rows) has been restored, and that the database was also rolled forward to the latest committed transaction in the test database.
The process described here takes only a few minutes, but the exact amount of time depends on the number of archived logs that need to be applied, which is a function of the frequency of the snapshot intervals and the redo rate of the database.