1 Amazon S3 Object Storage
Learn how Amazon S3 organizes object data, how its storage classes and lifecycle rules support different access patterns, and how to plan for data protection and recovery.
How Amazon S3 stores data
Amazon S3, or Amazon Simple Storage Service, is a scalable service. suits data that applications save and retrieve as complete files or data objects. S3 is a good fit when data is naturally handled as objects and accessed through S3 interfaces; workloads requiring a mounted shared file system or block storage for an operating system or database may need a different storage service.
Buckets, objects, and access
A is a container for objects, and you select its AWS Region when creating it. Each consists of its data and metadata, which can include content type or custom attributes. An 's key is its unique name within the . A key such as reports/2026/summary.pdf is a single name: the slash-separated parts look like folders but are part of the key.
Buckets and objects are private by default. Access can be granted through controls such as IAM policies and policies. S3 Versioning can keep multiple versions of an , which can help recover from some accidental overwrites or deletions.
Storage classes and access needs
S3 storage classes let you trade storage cost against access needs. Choose a class by considering how often data is accessed, how quickly it must be retrieved, how long it will be retained, and its resilience requirements. Retrieval charges and minimum storage durations may apply, so the lowest storage price does not always mean the lowest total cost.
S3 Standard: For frequently accessed data requiring rapid access, such as website assets, application data, or active analytics datasets.
S3 Intelligent-Tiering: For data with unknown or changing access patterns. It automatically moves eligible objects among access tiers. Optional archive tiers are available when retrieval taking minutes to hours is acceptable; monitoring and automation charges apply.
S3 Standard-IA: For long-lived, infrequently accessed data that is still needed with rapid access. Retrieval charges and a minimum storage duration apply.
S3 One Zone-IA: For infrequently accessed, re-creatable data when storing it in one Availability Zone is acceptable. It is not suited to data that must survive the loss of that zone.
S3 Glacier Instant Retrieval: For rarely accessed archive data that must be retrieved rapidly.
S3 Glacier Flexible Retrieval: For archive data where retrieval can take minutes to hours, such as some backup and disaster-recovery copies.
S3 Glacier Deep Archive: For long-term archives accessed very rarely, where retrieval taking hours is acceptable.
Automate storage changes
S3 Lifecycle rules can transition objects between eligible storage classes or expire them according to rules you define. They are useful when data becomes less active as it ages. For example, recent logs can remain readily available while older logs transition to archive storage.
and data protection
S3 Standard and several other multi-zone classes are designed for , often described as 11 nines. This is a design target for protecting stored objects against data loss, not a measure of whether the service can be accessed at a given time; that is availability.
alone does not prevent an authorized or accidental deletion from taking effect. For important data, consider protections such as versioning, retention controls, and a separate backup or recovery plan appropriate to the workload.
Common uses
S3 supports several common workloads:
Static website and application assets: Store images, downloads, and other objects that applications retrieve as needed.
Data lakes and analytics: Keep datasets in a central repository for analysis and machine-learning workflows.
Backups and disaster recovery: Store data copies, choosing a according to recovery-time needs.
Media and user-generated content: Store photos, video, and other uploaded files at scale.
Long-term archives: Retain records or historical data in Glacier classes when slower retrieval is acceptable.