Dell NetWorker 19.9 Administration Guide

Configuring checkpoint restart backups

The checkpoint restart feature allows a failed backup operation to restart at a known good point, before the point of failure during the backup.

A known good point is defined as a point in the backup data stream where the data is successfully written to the save set and that data can be located and accessed by subsequent recovery operations. This feature allows client backups that are part of a scheduled backup to be restarted, if they fail while running. This prevents the files and directories that have already been backed up from being backed up again.

Backup failures occur for various reasons. The most common reasons include hardware failures, loss of network connectivity, and primary storage software failures. If a backup fails and checkpoint restart is enabled, then failed save sets are marked as partial instead of as aborted. Partial save sets remain in the index, the media databases, and media such as AFTD.

You can manually restart a failed backup, or you can configure the backup to restart automatically. A restarted save set has a new SSID and savetime.

The NetWorker server and storage node components must remain running to manage the client failure and to create a partial save set. If the NetWorker server or storage node components fail during a backup, then partial save sets are not created. In this case, the backup for the checkpoint-enabled client starts from the beginning.

If the checkpoint restart feature is not enabled, a failure that is encountered during a scheduled backup operation might require a rerun of an entire backup tape set. This can be costly when a limited backup window of time is available, as a significant portion of the backup data might have been successfully transferred to tape, and the NetWorker software cannot resume a save set from the point of interruption.

For example, when performing an 800 GB backup that requires approximately 10 hours to complete and spans six tapes, if a failure occurs while writing to the last tape, the previous five tapes representing 9 hours of backup time may need to be rerun. As datasets continue to increase in size, so does the impact of backup failures.