# Amazon RDS Snapshot

> An RDS snapshot bills per GB-month of backup storage, and manual snapshots never expire on their own; snapshots of databases deleted years ago keep billing today. ZopNight discovers snapshots through a dedicated provider and recommends cleanup of orphaned and aged snapshots, the 2 categories that dominate backup waste.

Source: https://zop.dev/integrations/aws/rds-snapshot
Updated: 2026-08-19

---

An RDS snapshot is a point-in-time backup of a database instance or cluster, billed per GB-month of storage. Manual snapshots never expire on their own, so snapshots of long-deleted databases keep charging indefinitely.

## Automated backups end; manual snapshots do not

The distinction that drives RDS backup cost: automated backups live inside a retention window and delete themselves, with storage up to the database's provisioned size effectively included while the instance lives. Manual snapshots bill per GB-month with no expiry of any kind. Every manual snapshot is a decision to pay monthly forever, revocable only by deleting it, which nobody is ever assigned to do.

## The final-snapshot graveyard

Deleting an RDS instance prompts for a final snapshot, and prudence says yes. The database is gone; the snapshot remains, billing per GB-month with no owner, no expiry, and no database to restore into a running system anyone still operates. Multiply by every decommissioned environment over several years and final snapshots become a stratum of the bill: backups of things that no longer exist, kept by default rather than by intent.

## Pre-change snapshots pile on

The other accumulator is the safety snapshot: taken before a migration, an upgrade, a risky deploy. The change succeeds, the snapshot stays. Individually small, collectively these outnumber every other kind, and names like final-before-upgrade or pre-migration-2 date them precisely without ever triggering deletion.

## A dedicated discovery pass

RDS snapshots are discovered through a dedicated snapshot provider rather than the general resource sweep, with storage cost from Cost Explorer or CUR 2.0. Recommendations separate orphaned snapshots, where the source database no longer exists, from aged ones, because the safe-to-delete argument differs. Nothing schedules; a snapshot has no running state.

## Reviewing snapshots in the console

RDS console, then Snapshots. The Manual tab is where the graveyard lives; sorting by creation date puts the oldest, least defensible storage at the top.