Persistent Disk Snapshot

A snapshot is a point-in-time backup of a Persistent Disk, billed per GB of stored snapshot data per month. Snapshot schedules without retention policies grow storage costs indefinitely.

Incremental blocks, cumulative charges

Snapshot billing follows stored snapshot data, not the size of the source disk. The first snapshot of a disk stores its used blocks; each subsequent one stores only blocks that changed since the last. That incremental design keeps any single snapshot cheap and makes the whole chain deceptive: a daily schedule against a busy database disk adds changed blocks every day, and a chain that is never pruned compounds month over month. Deleting one snapshot in the middle does not necessarily free its full size either, because blocks still referenced by neighbours are retained.

Parented to the disk they protect

ZopDev discovers snapshots via Cloud Asset Inventory and links each one to its source disk as a child resource, so snapshot storage rolls up under the disk and workload it belongs to instead of floating unattributed. Spend comes from billing actuals, and recommendation rules flag aged snapshots. Snapshots whose source disk has since been deleted are split out into a separate orphaned-snapshot resource, a different problem with its own page. There is nothing to schedule on a snapshot; the levers are retention policy and deletion.

Retention gaps that let chains sprawl

Two patterns account for most snapshot waste. Schedules created without a retention rule, where an hourly or daily cadence quietly accumulates hundreds of snapshots per disk because nothing ever expires them. And manual pre-change snapshots, taken before a risky migration or upgrade, then forgotten the moment the change succeeds, surviving for years as insurance against a rollback nobody will ever perform.

Reviewing snapshot inventory in the console

Google Cloud console → Compute Engine → Snapshots lists every snapshot with its source disk, creation time, and size. Sorting by creation date surfaces the old tail fast; any snapshot schedule attached to a disk shows its retention setting under Compute Engine → Snapshots → Snapshot schedules, where a missing auto-delete rule is the thing to fix first.