S3 Standard: The Obvious Choice for Hot Data
This is the default storage class. S3 Standard is designed for frequently accessed data (Hot Data) that requires millisecond latency. Data is redundantly stored across at least 3 Availability Zones (AZs) within the same region, providing 99.999999999% durability (the famous “11 nines”).
Use cases: Static websites, mobile applications, content delivery (CDNs), real-time analytics.
Pricing model: The cost per GB of stored data is the highest, but API request (GET/PUT) and data retrieval charges are the lowest.
S3 Intelligent-Tiering: AI-Powered FinOps Automation
Manually managing the lifecycle of millions of files is a nightmare. S3 Intelligent-Tiering was designed for data with unknown or unpredictable access patterns, such as Data Lakes and user-generated content. This storage class monitors your files and automatically moves them between access tiers (Frequent, Infrequent, Archive) based on actual usage.
Use cases: Environments where the cost of human analysis exceeds the expected FinOps savings.
Pricing model: You pay a small monthly monitoring fee (a few cents per 1,000 objects), but there are no retrieval charges when data moves back to the hot tier. One important caveat: avoid using this class for millions of very small objects, as monitoring costs can quickly add up.
S3 Standard-IA (Infrequent Access): For Warm Data
This class is intended for data that is accessed less than once a month but still requires immediate access within milliseconds when needed (Warm Data). Resilience remains high, with data stored across 3 AZs.
Use cases: Secondary backups, synchronization files, older administrative documents.
Pricing model: The storage cost per GB is approximately 40% lower than S3 Standard. However, retrieval fees apply per GB read, the minimum storage duration is 30 days, and a minimum billable object size of 128 KB applies to each file.
Master and secure your cloud storage classes
tamper helps you audit, classify, and protect your S3 assets, completely agentless and with zero latency on any provider.
S3 One Zone-IA: Trading Redundancy for Savings
This is the only storage class, apart from Express, that stores data in a single Availability Zone (one geographic data center). It retains fast access and durability within that facility, but if a fire or major incident destroys the entire AZ, the data is permanently lost.
Use cases: Easily reproducible data, such as image thumbnails, or secondary backups where the primary data is stored on-premises.
Pricing model: 20% cheaper than Standard-IA. The same retrieval charges, minimum storage duration of 30 days, and minimum billable object size of 128 KB apply.
S3 Glacier Flexible Retrieval (Formerly Glacier): Cold Archives
This is where asynchronous storage begins (Cold Data). With Glacier, data is not available instantly. To read a file, you must first initiate a restoration request, which can take anywhere from a few minutes in expedited mode to 12 hours in standard or bulk mode.
Use cases: Legal archives, completed large-scale research datasets, long-term database backups.
Pricing model: The storage cost per GB is extremely low, approximately 70% cheaper than S3 Standard. The minimum billable storage duration increases to 90 days. Per-GB restoration charges apply, depending on the retrieval speed selected.
S3 Glacier Deep Archive: The Data Graveyard
This is AWS’s lowest-cost storage class, designed to replace traditional magnetic tapes (LTO) managed by physical storage providers. Data remains deeply dormant. Restoration takes at least 12 hours, and can take up to 48 hours for large-volume requests. Any access requirements must be anticipated.
Use cases: Regulatory compliance requiring logs or transactions to be retained for 5, 10, or 15 years, with no intention of accessing them again.
Pricing model: This is among the cheapest storage options in the cloud, at nearly one-tenth the price of S3 Standard. In return, the minimum billable storage duration is 180 days. Any file deleted before that period expires will still be charged for the full six months.
Note: For completeness, it is worth mentioning the more recently introduced S3 Express One Zone, a premium, ultra-fast storage class designed for AI workloads and high-performance computing. It bills by the minute but falls outside the scope of traditional storage tiering.
Conclusion
Managing S3 storage classes is a balancing act between cost per gigabyte, access charges (API requests and bandwidth), and technical constraints (restoration times and minimum object sizes). Properly orchestrating these classes through Lifecycle Policies is a hallmark of organizations that have reached cloud operations maturity.