Lifecycle Rules in Practice

A lifecycle rule consists of one or more actions applied to objects matching an optional filter (prefix and/or tags), triggered by the object's age in days since creation (or since becoming a non-current version, for versioned buckets). Transition actions move an object to a cheaper storage class at a specified day threshold; expiration actions delete the object entirely at a specified threshold.

A typical rule chains multiple transitions: Standard → Standard-IA at 30 days → Glacier Flexible Retrieval at 90 days → expire at 2555 days (7 years). AWS enforces minimum object age requirements between certain transitions (e.g. an object must be in Standard-IA for at least 30 days before transitioning further) to prevent lifecycle rules from causing more retrieval-fee churn than storage savings on objects that don't actually follow the assumed access pattern.

For versioned buckets, separate lifecycle actions can target 'noncurrent versions' specifically — e.g. expire non-current versions after 90 days, keeping only the current version long-term, which is essential to prevent Versioning's storage cost from growing unbounded as objects are repeatedly overwritten.