Amazon S3 Glacier Archive Strategy - Choosing a Storage Class and Retrieval Options
This article clarifies the criteria for choosing among Instant Retrieval / Flexible Retrieval / Deep Archive and introduces automatic tiering with lifecycle policies and compliance support with Vault Lock.
Overview of the Glacier Storage Classes
S3 Glacier is low-cost storage for long-term archiving, and it has three classes. Glacier Instant Retrieval provides millisecond access and is up to 68% cheaper than S3 Standard-IA. It is suited to archives of medical images and news photos accessed about once a quarter. Glacier Flexible Retrieval (formerly S3 Glacier) provides three retrieval options: expedited retrieval (1-5 minutes), standard retrieval (3-5 hours), and bulk retrieval (5-12 hours). It is suited to backup data accessed one to two times a year. Glacier Deep Archive is the lowest-cost storage class, and standard retrieval takes 12 hours and bulk retrieval takes 48 hours. It is suited to compliance data that regulatory requirements say must be retained for 7-10 years.
Automatic Tiering with Lifecycle Policies
With S3 lifecycle policies, you can automatically transition objects between storage classes based on the number of days since they were created. A typical design is a policy that transitions objects to Standard-IA 30 days after creation, to Glacier Flexible Retrieval at 90 days, and to Deep Archive at 365 days. Lifecycle policies can be filtered by prefix or tag, so you can manage the objects in a bucket with different policies for each type. When you enable the Archive Access tier and the Deep Archive Access tier of S3 Intelligent-Tiering, objects are automatically moved to Glacier-equivalent tiers based on access patterns. Unlike day-based lifecycle policies, the decision is based on actual access frequency, so it is suited to data with access patterns that are hard to predict.
Compliance and Vault Lock
Glacier Vault Lock (a feature of the original Amazon Glacier service that uses vaults; this service has stopped accepting new customers, announced on October 7, 2025) is a feature that applies a WORM policy to a vault. It prevents data from being deleted or changed once written. You can build tamper-proof archives that meet financial regulations such as SEC Rule 17a-4 and FINRA. Because a Vault Lock policy cannot be changed once it is locked, it is recommended that you check its behavior during the 24-hour validation period before completing the lock. S3 Object Lock also provides a similar WORM feature, but it allows control at the object level, and you can choose between Governance mode (privileged users can remove it) and Compliance mode (no one can remove it).
Glacier Pricing Comparison
Glacier Instant Retrieval costs about $0.004 per GB per month and allows retrieval in milliseconds (as of August 2026). Glacier Flexible Retrieval costs about $0.0036, with standard retrieval (3 to 5 hours) at about $0.01 per GB and expedited retrieval (1 to 5 minutes) at about $0.03. Glacier Deep Archive costs about $0.00099 and is the lowest-cost class, but standard retrieval takes 12 hours. The minimum storage duration is 90 days for Instant Retrieval, 90 days for Flexible Retrieval, and 180 days for Deep Archive, and early deletion is charged for the remaining period. When you set up automatic tiering with lifecycle policies, a design that takes the minimum storage duration into account is important.
Designing for Retrieval Costs and Lead Times
While Glacier has a low storage unit price, retrieving data takes time and money. In archive design, it is important to factor in not only storage costs but also the lead time and retrieval fees for when you actually need to retrieve data. The method of retrieving data quickly in an emergency is fast but relatively expensive, and the method of retrieving data in bulk over time is inexpensive. You choose a class based on the frequency of restores and the wait time you can tolerate. Retrieving a large amount of data at once drives up costs, so if you estimate in advance how often you might restore what, you can keep unexpected expenses down.
Using Glacier for Backup and Disaster Recovery
Glacier is suited to being the storage destination for backups and disaster recovery data that need long-term retention. You can store data that is not referenced day to day but must be retained for regulatory or business continuity reasons securely at low cost. You combine important data with replication to another Region to ensure geographic redundancy. If you integrate with a backup management service, you can automatically move data to archive tiers according to retention policies and centralize management up to and including generation management. It is essential to check recovery procedures regularly and verify that you can restore data within the expected time when it really matters.
Cautions on Minimum Storage Duration and Transition Costs
Glacier-family classes have a defined minimum storage duration, and if you delete data earlier than that, the charge for the remaining period is billed as an early deletion fee. If you casually move data that you might delete within a short period to a deep archive tier, it can end up costing more. When you design automatic tiering with lifecycle rules, you take this minimum storage duration into account and decide the timing of transitions carefully. Moving data between tiers itself also incurs request charges, so when you transition a large number of small objects, you keep costs down with measures such as handling them together.
Identifying Access Patterns
The key to choosing the right class is correctly identifying how often data is accessed and how long a retrieval time you can tolerate. If you might occasionally need to reference data immediately, you choose a tier that can retrieve it in milliseconds. If you touch the data only a few times a year and can wait several hours, a less expensive tier is suitable, and if you hardly ever reference it and only keep it for the long term, the lowest-cost tier is suitable. For data whose access patterns are hard to read, a mechanism that automatically moves it between tiers according to usage is convenient. Organizing requirements for each type of data and assigning a class that balances storage costs and retrievability makes for the optimal archive strategy.
Summary
Glacier is archive storage that stores infrequently accessed data at extremely low cost. You use the three classes, Instant Retrieval (retrieval in milliseconds), Flexible Retrieval (retrieval in minutes to hours), and Deep Archive (retrieval in 12 hours), according to your requirements for access frequency and retrieval time. You set up automatic transitions with S3 lifecycle rules and use Vault Lock or S3 Object Lock for WORM requirements.
References (Official AWS Resources)
The primary sources for this page are the official AWS website and documentation. Check the official pages below for the latest specifications and pricing.
If this page and the official documentation disagree, treat the official documentation as authoritative.