15.11 - Incomplete Backups - Data Stream Architecture

Teradata Data Stream Architecture (DSA) User Guide

Data Stream Architecture
December 2016
User Guide

The integrity of the database data is compromised if any part of the incremental backup is lost or corrupted. If any delta, cumulative, or full image required for a restore is missing or corrupt, a restore from any dependent backup image fails.

Incomplete backups are not subject to this limitation. An incomplete backup occurs if any incremental backup completed with errors, or was aborted and not re-run. In the event of a failed backup, prior and subsequent incremental backups are not affected. Similarly, when a backup completes with non-fatal errors, prior and future incremental backups do not use the backup image that received an error. Instead, subsequent incremental backup jobs use the most recent successful backup as the base. Therefore, it is important to fix the underlying cause of any error that occurs during incremental backups, and to re-run the incremental backup at the next available opportunity.

The following other situations may require that a new full backup be generated before any further delta or cumulative backups be run:
  • The system has gone through SYSINIT and/or a full database container (DBC) restore since the most recent full backup
  • The system has had an access module processor (AMP) reconfiguration or rebuild since the most recent full backup
  • The object list in the backup has been changed
  • The dictionary or data phase in the backup job has been changed
  • The target group in the backup job has been changed
  • Check retention invalidates or removes the last full save set in the DSC repository
  • The backup job is in the NEW state
The following backup jobs are always run as a full backup:
  • Any DBC only backup job
  • Any backup job in dictionary phase