Skip to main content

Amazon S3

  • What it is
    • Object storage: data stored as objects (file + metadata + key) inside buckets, retrieved over HTTP APIs.
    • Virtually unlimited capacity; a single object can be up to 5 TB.
    • Regional service, 11 nines (99.999999999%) of durability by replicating across multiple AZs.
  • Not a file system
    • No hierarchy — “folders” are just key prefixes. You cannot mount it as a disk or edit an object in place; you replace it.
  • Storage classes (know the trade-offs)
    • S3 Standard — frequent access, highest storage cost, no retrieval fee.
    • S3 Intelligent-Tiering — automatically moves objects between tiers based on access. Small monitoring fee. Best when access patterns are unknown.
    • S3 Standard-IA — infrequent access, cheaper storage, retrieval fee, still multi-AZ.
    • S3 One Zone-IA — same but stored in one AZ only. ~20% cheaper, lost if that AZ is destroyed. Only for reproducible data.
    • Amazon S3 Glacier family — archival, cheapest storage, retrieval measured in minutes to hours.
    • S3 Express One Zone — single-digit-millisecond, single-AZ, for high-frequency access.
  • Key features
    • Versioning, lifecycle policies to transition or expire objects, replication (cross-Region and same-Region).
    • Encryption at rest is on by default (SSE-S3); can use AWS Key Management Service (KMS).
    • Block Public Access is on by default — data is private unless you explicitly open it.
  • Not to be confused with

Linked from