Backups & restore
Automatic backup schedules, on-demand backups, point-in-time recovery, retention, and restoring.
Cloud backups are enabled by default when you provision a cluster. A cluster's Backups tab shows its schedule, its backups, and restore options.
Schedules
A backup schedule is defined by a snapshot time (UTC) and a set of retention tiers — at most one per frequency:
| Frequency | Default | Retention |
|---|---|---|
| Hourly | every 6 hours | 7 days |
| Daily | 02:00 UTC | 7 days |
| Weekly | Sunday, 02:00 UTC | 4 weeks |
| Monthly | 1st of the month, 02:00 UTC | 12 months |
Point-in-time recovery (PITR)
PITR requires backups to be enabled and an hourly retention tier. When enabled, you choose a restore window of 1, 2, 3, 5, or 7 days.
Some clusters use a platform-managed schedule (pgBackRest). These are shown read-only today — direct editing is coming later.
On-demand backups
If backups are enabled, take a backup at any time with Create backup. Choose a retention: 7 days (default), 1 day, 30 days, 90 days, or Until deleted (kept until you remove it manually, still bounded by the cluster's repository retention). Deleting a backup is permanent.
Restoring
Restore from any completed backup via Restore from backup (requires at least one backup to exist). The workspace-wide Backups view lists every cluster's last, next, and oldest snapshots and whether PITR is on.
Backups of deleted clusters: deleting a cluster with Keep backups for restore later checked keeps its backup data and removes its schedule (no future snapshots). Restoring from those retained backups after the cluster is gone is not supported yet; restore works for clusters that still exist.