Data retention answers three practical questions: which data must be kept, how long it must be kept, and what happens when that period ends. A policy should identify the data owner, the reason for retention, the event that starts the retention clock, and any exceptions. For example, operational logs might be retained for 90 days, while financial records might need several years of retention under applicable requirements. These periods are organizational decisions, not universal cloud defaults.
Retention has both minimum and maximum boundaries. A minimum period prevents premature destruction of required evidence or recovery data. A maximum period limits unnecessary storage, privacy exposure, and the impact of a security incident. Keeping everything forever is not automatically safer: it increases cost and can conflict with data minimization or deletion obligations. Legal holds can suspend normal disposal, so deletion workflows must account for them.
Cloud lifecycle policies automate actions based on conditions such as object age, tags, or access patterns. Moving data to a cheaper storage tier changes its storage characteristics, not necessarily its retention period. For example, an application could retain logs in readily accessible storage initially, transition older logs to archival storage, and expire them after the approved retention period. Engineers must consider retrieval latency, retrieval charges, minimum storage duration charges, and transition eligibility before choosing tiers.
Amazon S3 Lifecycle can transition objects between supported storage classes and expire objects. In a versioned bucket, expiration of a current object generally creates a delete marker rather than permanently removing its previous versions. Noncurrent-version expiration requires separate configuration. Rules must therefore address the relevant versions and copies; hiding an object from normal reads does not prove that its data has been deleted.
Retention enforcement differs from automated cleanup. S3 Object Lock protects individual object versions using retention periods or legal holds. Governance mode permits authorized bypass, whereas compliance mode prevents deletion or shortening of the retention period even by the account root user during that period. A legal hold has no automatic expiration and remains until removed by an authorized principal. Object Lock requires versioning, and lifecycle expiration cannot permanently delete a protected version while its protection remains active.
Backups have their own retention requirements. AWS Backup provides centralized scheduling and retention management for supported resources, helping preserve recovery points after production data changes or is deleted. Backup frequency determines how much recent work might be lost; backup retention determines how far back recovery can reach. Neither a successful backup job nor a long retention period proves that restoration will meet operational requirements, so recovery testing remains necessary. AWS Backup Vault Lock can enforce retention constraints and protect recovery points against premature deletion, depending on its configuration and mode.
File storage illustrates another important distinction. Amazon EFS Lifecycle management moves eligible files among storage classes according to access patterns; it does not provide age-based automatic file deletion. Amazon FSx for Lustre supports automatic backup scheduling and retention for supported persistent file systems, while scratch file systems do not support backups. File disposal and backup expiration therefore require separate design decisions.
A complete implementation covers production data, replicas, exports, object versions, and recovery points. Retained encrypted data also requires usable encryption keys and appropriate access permissions throughout the retention period. Monitor policy changes, failed backups, and deletion outcomes, and document how exceptions are approved. For exam questions, distinguish tiering for cost, expiration for disposal, backups for recovery, and immutable retention for protection against deletion.
Amazon S3
AWS Backup