SafeGrd vs pgBackRest
pgBackRest is the open-source standard for physical PostgreSQL backups. SafeGrd takes logical backups of PostgreSQL and more, restores them on a schedule to test them, and runs from one install command with a console and alerts.
| Feature | SafeGrd | pgBackRest |
|---|---|---|
| Scheduled restore tests | Yes | No |
| Tables and row counts checked after restore | Yes | Checksums only |
| Console, alerts and a status badge | Yes | No |
| Encrypted by default | Yes | Off by default |
| Backups written under Object Lock | Yes | No |
| Each backup uploads only what changed | Yes | Yes |
| Restore one table into a running database | Yes | No |
| Restores into a newer PostgreSQL major version | Yes | No |
| MySQL, MongoDB, SQLite, files, mailboxes, WordPress sites and Laravel apps | Yes | No |
| Signed record you can verify offline | Yes | No |
| A locked snapshot before an AI agent’s destructive command | Yes | No |
| Restore tests run for you, with no server of your own | Yes | No |
| Point-in-time recovery | Self-hosted PostgreSQL and SQLite | Yes |
pgBackRest is the better fit when the database is too large to dump: its physical backups
restore in parallel and by delta. Both need archive_command for
point-in-time recovery, so neither gives it on a managed database where you cannot set it
(SafeGrd's). Running both works well: SafeGrd needs only a read-only role and can read
from a replica (Databases), which gives you a
tested logical copy, with the schema's roles, owners and grants, next to pgBackRest's
physical one.
Why teams add SafeGrd
pgbackrest verifychecks file checksums. It does not restore. SafeGrd restores the newest backup and counts every row (Restore and Fire Drills).- A logical copy loads into an empty database on any server, which a physical backup of
the cluster's files cannot. One table loads into a database that is running, with
restore --table(Databases). - Each PostgreSQL backup uploads only the chunks of each table that changed, and can skip reading tables nobody wrote (the change log). Between two backups, a self-hosted PostgreSQL restores to any second through its WAL (point-in-time recovery).
- One place for PostgreSQL and everything else you run (Surfaces).
pgBackRest details come from its documentation and repository.