Database Backup & Recovery
Snapshot frequency, retention, point-in-time restore, and geo-redundancy.
This document describes our production backup policy: backup frequency, retention, snapshot timing, point-in-time restore, and cross-region redundancy. The goal is a recoverable database with predictable restore windows.
Backup policy time
- Snapshot time (UTC): 20:00
- Local time zone: GMT+5:30 → 1:00 AM IST
- This snapshot time applies uniformly across all configured backup frequency units.
Frequency and retention
Three backup frequency units are configured, each with its own retention policy. All use the global snapshot time above.
| Frequency unit | Interval | Retention | Snapshot time |
|---|---|---|---|
| Hourly snapshot | Every 6 hours | 7 days | 1:00 AM IST |
| Weekly snapshot | Every Saturday | 4 weeks | 1:00 AM IST |
| Monthly snapshot | Last day of the month | 12 months | 1:00 AM IST |
Point-in-time restore (PITR)
- Restore window: 7 days.
- Data can be restored to any point in time within the past 7 days.
Additional backup copies (geo-redundancy)
Backups are copied to other regions for better safety and quicker restore if one region fails.
| Region | Role |
|---|---|
| Pune (centralindia) | Original backup region |
| Chennai (southindia) | Stores copies of hourly snapshots |
Related
- Encrypt backups and snapshots with KMS — see Encryption at Rest.
- Enable continuous backups and delete protection — see Production Best Practices.