Amazon S3 offers several storage classes designed to meet various needs for durability, availability, and cost-efficiency. Each storage class is optimized for a different use case, whether that be for frequently accessed data, archival storage, or long-term backups.
S3 Standard
Use Case: Frequently accessed data that requires low latency and high throughput.
Durability: 99.999999999% (11 9’s)
Availability: 99.99% (four 9’s)
Performance: Low latency and high throughput.
Cost: Higher storage cost compared to other classes, but ideal for frequently accessed data.
Example Use Cases:
- Active websites
- Big data analytics
- Content distribution
S3 Standard-IA (Infrequent Access)
Use Case: Data that is infrequently accessed, but still requires rapid access when needed.
Durability: 99.999999999% (11 9’s)
Availability: 99.9% (three 9’s)
Performance: Low latency and high throughput.
Cost: Lower storage cost than S3 Standard, but retrieval costs are higher.
Minimum Object Size: 128 KB (otherwise, there are additional fees).
Example Use Cases:
- Backup data
- Disaster recovery
- Long-term storage of data that’s rarely accessed
S3 One Zone-IA
Use Case: Infrequently accessed data that can be recreated if lost, but needs to be stored cost-effectively.
Durability: 99.999999999% (11 9’s), but only within a single Availability Zone.
Availability: 99.5% (two 9’s)
Performance: Low latency and high throughput.
Cost: Lowest cost for infrequently accessed data, but data is stored in only one Availability Zone (reduces durability compared to other IA classes).
Example Use Cases:
- Secondary backups
- Data that is not critical or can be easily replicated if lost
- Storing data for a limited time period before deletion
S3 Glacier
Use Case: Archival data that is rarely accessed, but must be retained for long periods (e.g., compliance or regulations).
Durability: 99.999999999% (11 9’s)
Availability: 99.99% (four 9’s)
Performance: Retrieval times vary from minutes to hours.
Cost: Very low storage cost, but retrieval costs can be high.
Retrieval Time:
Expedited: Available within 1-5 minutes.
Standard: Available within 3-5 hours.
Bulk: Available within 5-12 hours.
Example Use Cases:
- Archival of logs
- Long-term backup
- Compliance data storage
S3 Glacier Deep Archive
Use Case: Lowest-cost storage for data that is rarely accessed and is kept for long periods, with retrieval times that are less time-sensitive.
Durability: 99.999999999% (11 9’s)
Availability: 99.9% (three 9’s)
Performance: Retrieval times range from 12 hours to 48 hours.
Cost: The lowest storage cost of all S3 storage classes.
Retrieval Time:
Standard: 12 hours
Bulk: 48 hours
Example Use Cases:
- Long-term data archival (e.g., 7-10 years or more).
- Compliance data that must be kept for extended periods but rarely accessed.
- Large-scale data storage that is archived, such as scientific data or records.
S3 Intelligent-Tiering
Use Case: Data with unknown or changing access patterns, where it’s difficult to predict which objects will be accessed frequently or infrequently.
Durability: 99.999999999% (11 9’s)
Availability: 99.9% (three 9’s)
Performance: Low latency and high throughput.
Cost: Cost-efficient, as it automatically moves data between two access tiers: frequent access and infrequent access, depending on usage.
No Minimum Object Size: Unlike other storage classes like S3 Standard-IA, S3 Intelligent-Tiering doesn’t impose a minimum object size, making it more flexible for varying types of data.
Classes:
- Frequent Access Tier: For objects that are accessed more frequently
- Infrequent Access Tier: For objects that are accessed less frequently
- Archive Access Tier: For objects that are rarely accessed but still need to be available for occasional retrieval (also known as archive data).
- Deep Archive Access Tier: For objects that are very rarely accessed and can tolerate long retrieval times (ideal for archival data with minimal retrieval requirements).
Tier | Use Case | Storage Cost | Retrieval Cost | Minimum Object Size | Access Times | Transition Rules |
---|---|---|---|---|---|---|
Frequent Access | For objects that are accessed frequently. | Highest storage cost | Standard retrieval costs | None | Instant access | Moves to Infrequent Access after 30 days of no access. |
Infrequent Access | For objects that are accessed less frequently. | Lower storage cost | Higher retrieval costs | None | Instant access | Moves back to Frequent Access after access within 30 days. |
Archive Access | For rarely accessed data that may require occasional retrieval (e.g., backup data). | Lower storage cost than Frequent Access | Higher retrieval costs (within hours) | None | 3-12 hours | Moves from Frequent or Infrequent after 90 days of no access. |
Deep Archive Access | For very rarely accessed data that is retained for long-term archive. | Lowest storage cost | Very high retrieval costs (12-48 hours) | None | 12-48 hours | Moves from Frequent or Infrequent after 180 days of no access. |
Example Use Cases:
- Data with unpredictable or changing access patterns.
- Data that might be accessed for short bursts, but not continuously.
Comparison
Storage Class | Use Case | Durability | Availability | Cost | Retrieval Times |
---|---|---|---|---|---|
S3 Standard | Frequently accessed data | 99.999999999% | 99.99% | Highest storage cost | Instant access |
S3 Intelligent-Tiering | Unknown or changing access patterns | 99.999999999% | 99.9% | Medium cost | Instant access |
S3 Standard-IA | Infrequently accessed data | 99.999999999% | 99.9% | Lower than Standard | Instant access |
S3 One Zone-IA | Infrequent data that can be recreated if lost | 99.999999999% | 99.5% | Lowest for IA data | Instant access |
S3 Glacier | Archival data with rare access needs | 99.999999999% | 99.99% | Very low | 1-12 hours |
S3 Glacier Deep Archive | Extremely low-cost archival data with minimal access | 99.999999999% | 99.9% | Lowest storage cost | 12-48 hours |
S3 Outposts | Data storage in on-premises environments (hybrid) | 99.999999999% | Varies | Based on Outposts | Instant access |
Choosing the Right S3 Storage Class:
Frequent Access: Use S3 Standard for data that is actively used and requires fast access.
Infrequent Access: Use S3 Standard-IA or S3 One Zone-IA for data that is accessed less frequently but still needs to be readily available.
Archival Data: Use S3 Glacier for data that needs to be archived with relatively fast access (3-5 hours).
Long-Term Archive: Use S3 Glacier Deep Archive for data that is rarely accessed but must be retained for long periods (12-48 hours retrieval).
Dynamic Access Needs: Use S3 Intelligent-Tiering for data where access patterns are unpredictable.
