CSE Roadmap

Core tier · about 2 hours · 5 parts

Part 4 of 5 · 2 min read

Why a backup must live on separate media

A backup stored on the same disk as the live database offers no protection against the exact failure it exists to protect against.

Why a backup must live on separate media

Think of keeping the only backup copy of an important document in the very same drawer as the original. That does nothing at all to protect against the drawer itself being destroyed.

In plain words: as covered on this level's first page, a disk failure can destroy both the data files and the write-ahead log if they are stored on the same physical disk. A backup stored on that same disk offers no protection against exactly this failure, since the disk failure would take the backup down right along with everything else it was meant to protect. A backup must be stored on genuinely separate physical media, ideally in a separate physical location entirely, for it to actually serve its purpose when the specific failure it exists to guard against actually occurs.

A worked example. A backup file saved to the same disk as the live database provides no real protection if that disk fails, since the backup is destroyed along with the data it was supposed to help recover.

The key lines. No new SQL; backup storage location is an operational decision, not something expressed in a backup command itself.

Backup storage separation

Backup storage separation is the requirement that a backup be stored on genuinely separate physical media from the live database, since a backup on the same disk provides no protection against the disk failure it exists to guard against.

Example: keeping a document's only backup copy in the exact same drawer as the original, offering no protection if that drawer is destroyed, mirrors a backup stored on the same disk.

Remember this

Crash recovery, covered earlier in this level, handles a system crash using the log alone, with the database still running on the same disk. Backups, combined with point-in-time recovery, handle disk failure and human mistakes, which require a separately stored copy entirely. Each tool matches a different failure from the first page's classification.

Quick notes

  • A full backup copies everything; an incremental backup copies only changes since the previous backup.
  • Point-in-time recovery restores a backup, then replays the log forward to a specific chosen moment.
  • Point-in-time recovery is specifically useful for recovering from a human mistake, not just a disk failure.
  • A backup stored on the same disk as the live database offers no protection against a disk failure.
  • Backups and crash recovery are separate tools, matched to separate failure types from this level's first page.