Amazon S3 Glacier
Low-cost archival storage for long-term data retention
Glacier is like a deep freeze for your data; it's incredibly cheap storage for stuff you rarely need but must keep for years (think legal documents, old backups, compliance records). The trade-off? Retrieving data takes time, anywhere from minutes to hours, depending on how much you're willing to pay. It's like storing boxes in a warehouse basement: super cheap rent, but when you need something, it takes a while to dig it out. Perfect for 'write once, read never (or almost never)' scenarios. You're trading access speed for massive cost savings of up to 95% cheaper than regular S3 storage.
Amazon Glacier (now part of S3 Glacier storage classes) provides ultra-low-cost archival storage with retrieval times ranging from minutes to hours. S3 Glacier offers three storage classes: S3 Glacier Instant Retrieval (millisecond access, for quarterly access), S3 Glacier Flexible Retrieval (minutes to hours, for annual access), and S3 Glacier Deep Archive (12-48 hours, for 7-10 year retention). The original vault-based Glacier service (the standalone Glacier API) is no longer accepting new customers, so new workloads should use the S3 Glacier storage classes and address archives as ordinary S3 objects.
Key Capabilities
- Offers three retrieval tiers: Expedited (1-5 minutes), Standard (3-5 hours), and Bulk (5-12 hours), with cost scaling inversely with speed
- S3 Glacier Instant Retrieval provides millisecond access to archived data while still delivering low archival storage costs
- Integrates with S3 lifecycle rules to automatically archive objects from Standard or IA storage classes after a defined number of days
- S3 Object Lock enforces WORM (Write Once Read Many) compliance policies that are immutable once applied, supporting regulatory retention requirements
- S3 Glacier Flexible Retrieval and Deep Archive carry a 90-day and 180-day minimum storage duration respectively; early deletion incurs a prorated fee
- Data is redundantly stored across multiple AZs within a region, providing the same durability guarantee (11 nines) as standard S3
Gotchas & Constraints
Gotcha #1: Glacier has minimum storage duration charges. Deleting data before 90 days (Flexible Retrieval) or 180 days (Deep Archive) incurs early deletion fees. Gotcha #2: Retrieval costs money: bulk retrievals are cheap but slow, expedited retrievals are fast but expensive. Constraints: Not suitable for frequently accessed data. Use S3 Intelligent-Tiering if access patterns are unpredictable.
A healthcare provider must retain patient records for 10 years per HIPAA regulations. They have 2PB of medical imaging data (X-rays, MRIs, CT scans) that's rarely accessed after 1 year but must be available if needed. Storing this in S3 Standard would cost $46,000/month. They implement a lifecycle policy: new data stays in S3 Standard for 30 days (frequent access during active treatment), transitions to S3 Glacier Flexible Retrieval after 30 days (occasional access for follow-ups), and moves to S3 Glacier Deep Archive after 1 year (long-term archival). Storage cost drops to $2,000/month, a 96% reduction. When a patient requests records from 5 years ago, they initiate a standard retrieval (12 hours, $0.03/GB), and the data is available the next day. They use S3 Object Lock in compliance mode to enforce immutability: records can't be deleted or modified for 10 years, ensuring compliance.