There are two ways

  • manual (snapshots or via aws backups)
  • automated (1 - 35 days)

Recovery

  1. Automated
    1. AWS Console → RDS → “Restore to point in time”.
    2. Creates a new DB instance from the backup + logs.
  2. Manual
    1. Console → RDS → Snapshots → Select → “Restore Snapshot”.

    2. Creates a new DB instance from that snapshot.

    3. Backup console → Recovery points → “Restore”

    4. Also creates a new RDS instance.

When

  • Single AZ
    • Backup takes I/O
    • PITR and snapshots
  • Muliti AZ
    • Takes from standby instance, no I/O effects
    • PIRT and Snapshots
  • RR
    • Replicas aren’t used for backups, still taken from primary instance, partly because replication is async
    • Why would you back up a back up (kinda)
    • If you want, promote to master and take it from there