S3 Glacier Storage Classes: Archival Tiers
Amazon S3 Glacier offers specialized storage classes optimized for archival data, providing different access patterns, retrieval times, and cost structures. These tiers are crucial for long-term data retention with cost efficiency.
The Glacier family comprises three distinct storage classes designed for various archival needs, ranging from milliseconds to hours for retrieval.
Amazon S3 Glacier Instant Retrieval
Use Case: Archival data requiring fast access. This class is designed for frequently accessed archives with immediate retrieval needs.
latency
Millisecond latency access
availability_zone_replication
Data copied to >3 AZs
retrieval_fee
Per GB
minimum_storage_duration
90 days
minimum_object_size
128 KB
Use Cases:
- Archival data requiring fast access
Amazon S3 Glacier Flexible Retrieval
Use Case: Archival data, where retrieval time can vary. This class offers flexible retrieval options suitable for less urgent data.
availability_zone_replication
Data copied to >3 AZs
retrieval_modes
Expedited (1-5 minutes), Standard (3-5 hours), Bulk (5-12 hours)
retrieval_fees
Different fees for each mode
minimum_storage_duration
90 days
minimum_object_size
128 KB
Use Cases:
- Archival data, where retrieval time can vary
Amazon S3 Glacier Deep Archive
Use Case: Long-term storage for compliance, regulatory requirements, or infrequently accessed data. It is the lowest-cost storage option for archiving.
availability_zone_replication
Data copied to >3 AZs
access_modes
Standard (Within 12 hours), Bulk (Within 48 hours)
retrieval_fee
Based on retrieval speed
minimum_storage_duration
180 days
minimum_object_size
40 KB
Use Cases:
- Long-term storage (compliance, regulatory), infrequent access
Data Access Spectrum: S3 and Glacier Tiers Overview
The following lists the S3 and Glacier storage classes from most frequent access to archive, indicating their relative position in the data access spectrum.
The spectrum illustrates the progression from frequently accessed data to long-term archive storage, ordered by decreasing access frequency and typically decreasing cost per GB (excluding retrieval fees).
Technical Specs: Order: S3 Standard → S3 Intelligent Tiering → S3 Express One Zone (High Performance) → S3 Standard IA → S3 One Zone IA → S3 Glacier Instant Retrieval → S3 Glacier Flexible Retrieval → S3 Glacier Deep Archive
Amazon S3 Glacier is cheaper for long-term archival, but retrieval takes minutes to hours, not meeting millisecond access requirements for frequently accessed data.
Technical Specs: Retrieval time: minutes to hours (for S3 Glacier Flexible Retrieval and Deep Archive)
S3 Glacier Data Retrieval Methods
S3 Glacier provides specific mechanisms for data retrieval, including direct access for subsets of data and batch operations for restoring archives.
Glacier Select allows SQL queries on data stored in Glacier without restoring the entire archive, efficiently retrieving subsets of data.
Technical Specs: Retrieval Options: Expedited: 1-5 minutes, Standard: 3-5 hours, Bulk: 5-12 hours. Output: Results are placed in an S3 bucket for retrieval.
AWS announced discontinuation of S3 Select and Glacier Select features for new customers starting July 25, 2024. Existing customers can continue to use them. This topic remains valid for the certification.
S3 Batch Operations can be used to restore archive objects from S3 Glacier, allowing the same action to be performed on a massive number of S3 objects without complex scripting.
Glacier Vault Lock for Compliance
Glacier Vault Lock enforces Write Once, Read Many (WORM) protections at the vault level for data stored in S3 Glacier, crucial for strict regulatory compliance.
Glacier Vault Lock applies Write Once, Read Many (WORM) protections at the vault level for data stored in S3 Glacier.
Unlike S3 Object Lock which works on individual objects, Glacier Vault Lock works on the entire vault.
A WORM policy, such as a retention period, is defined for the vault.
Once the policy is locked, it becomes immutable and cannot be changed or deleted.
The locking process involves a two-step initiation with a 24-hour window to abort or make changes.
Technical Specs: 24-hour window to abort or make changes
The Glacier Vault Lock is permanent once the policy is immutable.
S3 Lifecycle Management with Glacier Tiers
S3 Lifecycle rules automate the movement of objects to and between Glacier storage classes to optimize costs and manage data retention policies.
Objects can be automatically transitioned from various S3 Standard and Infrequent Access tiers to Glacier storage classes. For example, from S3 Standard to Glacier Instant Retrieval, Glacier Flexible Retrieval, or Glacier Deep Archive; from S3 Standard IA to the same Glacier tiers; or from S3 One Zone IA to Glacier Instant Retrieval, Glacier Flexible Retrieval, or Glacier Deep Archive.
Transitions are also supported between Glacier tiers for further cost optimization. For example, from Glacier Instant Retrieval to Glacier Flexible Retrieval or Glacier Deep Archive; and from Glacier Flexible Retrieval to Glacier Deep Archive.
Objects must be stored for at least 30 days in S3 Standard IA or S3 One Zone IA before they can be transitioned out.
Technical Specs: Minimum storage duration: 30 days in S3 Standard IA or S3 One Zone IA before transition out
Transitioning objects out before their minimum storage duration incurs charges for the remainder of that duration.
Multiple transitions can be chained in a lifecycle rule, for example, moving data from S3 Standard to S3 Standard-IA after 30 days and then to Glacier Deep Archive after another 30 days.
Moving to S3 Glacier Deep Archive requires a minimum retention period after the transition to S3 Standard-IA; for instance, if transitioning to Standard-IA at day 30 and then to Glacier Deep Archive at day 60, the object must remain in Standard-IA for at least 30 days.
Technical Specs: Requires min 30 days retention in S3 Standard-IA before transitioning to S3 Glacier Deep Archive
S3 Event Notifications can trigger actions in real-time when specific events occur in an S3 bucket, including s3:ObjectRestore:*, indicating an object has been restored from Glacier.
Technical Specs: Event Type: s3:ObjectRestore:*
Glacier Integration with Other AWS Services
S3 Glacier integrates with various AWS services to facilitate data migration, backup, and hybrid cloud storage.
Tape Gateway maps Virtual Tape Library (VTL) to S3 or S3 Glacier. When ejecting a virtual tape, it can be automatically pushed to S3 Glacier or S3 Glacier Deep Archive for cost-effective long-term retention. Data is first staged in S3; Tape Gateway does NOT directly write to S3 Glacier or Deep Archive.
Technical Specs: Tape Gateway maps Virtual Tape Library (VTL) to S3 or S3 Glacier. Important Note: Tape Gateway does NOT directly write to S3 Glacier or Deep Archive; data is first staged in S3.
DataSync can write directly to an S3 bucket with Glacier Flexible Retrieval or Deep Archive as the configured storage class. This applies to S3-based Glacier storage classes only — not to legacy vault-based Glacier. Configuring the destination storage class in the DataSync task eliminates the need to transfer to S3 Standard first.
S3 Glacier Pricing Considerations
Understanding the pricing model for S3 Glacier is essential for cost optimization, particularly the storage charges and retrieval fees.
Archive tiers (Glacier) have lower per GB storage charges, making them cost-effective for long-term archiving.
S3 Glacier storage classes incur retrieval fees, which vary based on the retrieval mode selected (Expedited, Standard, Bulk).
Each Glacier storage class has a minimum storage duration, and objects deleted, overwritten, or moved before this period are still charged for the full duration.
Technical Specs: Minimum storage duration: 90 days for Glacier Instant Retrieval/Flexible Retrieval, 180 days for Glacier Deep Archive.