Storage tiering & lifecycle

Storage cost is mostly a function of how much data you keep and which tier it sits in. Data that is written once and rarely read still bills at hot-tier rates; snapshots and backups accumulate well past any retention policy; buckets and accounts have no lifecycle rule to age data down automatically.

What we look for

• Cold data on a hot tier: blobs/objects not read for a long window that belong in Cool/Cold/Archive (or Intelligent-Tiering / Autoclass).
• Over-retention: snapshots, point-in-time restore, long-term retention, soft-delete, or versioning kept far beyond the stated policy.
• Missing lifecycle: an account/bucket with no rule to tier or expire data, so it only ever grows.
• Registry bloat: a container registry holding untagged image manifests that no tag points to, which still bill as storage past the SKU’s included quota. A purge of the ones older than a set age frees the overage; a manifest pulled by digest is the one case to rule out first.

Retention changes are checked against your data-retention and compliance policy first: we never recommend shortening a window that an audit requirement needs.

How to act

A move to an automatic tier (S3 Intelligent-Tiering, GCS Autoclass) is reversible: it charges no retrieval fee, and the data can be copied back. A rule that moves data to a cold or archive tier is one-way: getting the data back means a restore with retrieval fees and a minimum stay. Deleting snapshots/backups is one-way too, so those candidates lead with a snapshot-or-export pre-step and a grace window. leancosts drafts the lifecycle rule or retention change and what to watch once the data has aged past the rule: about 35 days for Intelligent-Tiering, which bills an object at the Infrequent Access rate only after 30 days unread. Follow Act on a cost finding.