Skip to main content
All articles

Cloud Storage Tiering: When to Use Standard, Infrequent, Archive, and Deep Archive

A decision framework for storage tiering across S3, Azure Blob, Cloud Storage, and OCI Object Storage with lifecycle automation.

Jeff MonfieldFebruary 6, 20269 min read

Why Storage Tiering Matters

Object storage is the foundation of cloud data management. Every backup, log file, data lake, media asset, and application artifact lives in object storage. Left unmanaged, object storage costs grow linearly with data volume — and data volume only goes up. Most organizations store the vast majority of their data in the most expensive storage tier because it is the default, and they never configure lifecycle policies to move cold data to cheaper tiers. The result is paying premium hot-storage prices for data that has not been accessed in months or years.

Storage tiering solves this by providing multiple storage classes at different price points, each optimized for different access patterns. Hot storage provides instant access at the highest per-GB cost. Infrequent access storage reduces cost by 40-50 percent but adds a retrieval fee. Archive storage reduces cost by 70-90 percent but requires minutes to hours for data retrieval. Deep archive storage is the cheapest option at 90-95 percent savings but requires hours for retrieval. By matching your data to the appropriate tier, you can reduce storage costs by 50 to 80 percent without losing access to any data.

AWS S3 Storage Classes

S3 offers seven storage classes, from hottest to coldest. S3 Standard ($0.023/GB/month) is the default for frequently accessed data. S3 Intelligent-Tiering ($0.023/GB/month for frequent, $0.0125/GB for infrequent, $0.004/GB for archive instant access) automatically moves objects between tiers based on access patterns with no retrieval fees — this is the best default choice for data with unpredictable access patterns. S3 Standard-IA ($0.0125/GB/month) is for data accessed less than once per month, with a $0.01/GB retrieval fee and a 30-day minimum storage duration. S3 One Zone-IA ($0.01/GB/month) is like Standard-IA but stores data in a single AZ, reducing cost at the expense of reduced durability.

For archival data, S3 Glacier Instant Retrieval ($0.004/GB/month) provides millisecond retrieval with a $0.03/GB retrieval fee, ideal for data accessed once per quarter. S3 Glacier Flexible Retrieval ($0.0036/GB/month) provides retrieval in 1-5 minutes (expedited), 3-5 hours (standard), or 5-12 hours (bulk) with retrieval fees ranging from $0.01 to $0.03/GB. S3 Glacier Deep Archive ($0.00099/GB/month) is the cheapest option at under a penny per GB per month, with retrieval taking 12 hours (standard) or 48 hours (bulk) and costing $0.02/GB for standard retrieval.

A 10 TB dataset stored entirely in S3 Standard costs $230/month. Moving 8 TB of rarely accessed data to Glacier Instant Retrieval reduces the cost to $46 + $32 = $78/month — a 66 percent reduction. Moving the cold data to Glacier Deep Archive reduces it further to $46 + $8 = $54/month — a 77 percent reduction.

S3 Intelligent-Tiering recommendation

For buckets where you cannot predict access patterns (and that includes most buckets), enable S3 Intelligent-Tiering as the default storage class. The monitoring fee is $0.0025 per 1,000 objects — negligible compared to the savings. Intelligent-Tiering moves data automatically between frequent, infrequent, and archive tiers with no retrieval fees and no operational overhead.

Estimate S3 costs by storage classBuild S3 lifecycle rules for automatic tiering

Azure Blob Storage Tiers

Azure Blob Storage offers four access tiers. Hot ($0.018/GB/month in East US) is for frequently accessed data. Cool ($0.01/GB/month) is for data accessed infrequently, with a 30-day minimum storage duration and a $0.01/GB retrieval fee. Cold ($0.0036/GB/month) was introduced in 2024 for data accessed rarely, with a 90-day minimum storage duration and $0.03/GB retrieval fee. Archive ($0.00099/GB/month) is for long-term storage with retrieval requiring rehydration that takes up to 15 hours (standard) or 1 hour (high priority) at $0.02/GB.

Azure's tiering model has a unique feature: you can set the access tier at the blob level, not just the container or bucket level. This allows fine-grained tiering where frequently accessed objects in a container stay in Hot while older objects in the same container move to Cool or Cold. Azure also supports lifecycle management policies that transition blobs between tiers based on last modified date, last accessed date (with access tracking enabled), or creation date.

The Cold tier is particularly useful for data that sits between Cool and Archive in access frequency — data accessed a few times per year but that needs to be available within seconds when accessed. Before the Cold tier was introduced, teams had to choose between the higher cost of Cool and the hours-long retrieval time of Archive. Cold fills this gap at a price point close to Archive but with immediate access.

Estimate Azure Storage costs by tierBuild Azure Blob lifecycle policies

GCP Cloud Storage Classes

GCP Cloud Storage offers four storage classes. Standard ($0.020/GB/month in US multi-region, $0.023/GB for single region) is for frequently accessed data. Nearline ($0.010/GB/month) is for data accessed less than once per month, with a 30-day minimum storage duration and a $0.01/GB retrieval fee. Coldline ($0.004/GB/month) is for data accessed less than once per quarter, with a 90-day minimum storage duration and a $0.02/GB retrieval fee. Archive ($0.0012/GB/month) is for data accessed less than once per year, with a 365-day minimum storage duration and a $0.05/GB retrieval fee.

GCP's storage class pricing is competitive and straightforward. A key advantage is that all storage classes provide the same millisecond-latency access — there is no rehydration delay for any class, including Archive. The difference is the per-GB retrieval fee and the minimum storage duration. This means you can access archived data immediately, but you pay a higher retrieval fee. For workloads where occasional urgent access to archived data is needed, GCP's model is more convenient than AWS Glacier or Azure Archive, which require hours for retrieval.

GCP also offers Autoclass, which is similar to S3 Intelligent-Tiering. Autoclass automatically moves objects between storage classes based on access patterns. Objects start in Standard and are downgraded to Nearline after 30 days without access, then to Coldline after 90 days, and finally to Archive after 365 days. If an object is accessed, it is upgraded back to Standard. Autoclass simplifies tiering management for buckets with mixed access patterns.

Estimate Cloud Storage costsBuild Cloud Storage lifecycle rules

OCI Object Storage Tiers

OCI Object Storage offers three storage tiers. Standard ($0.0255/GB/month) is for frequently accessed data. Infrequent Access ($0.01/GB/month) is for data accessed less than once per month, with a $0.01/GB retrieval fee and a 31-day minimum storage duration. Archive ($0.0026/GB/month) is for long-term storage with a minimum 90-day storage duration and retrieval requiring a restore request that takes approximately 1 hour, plus $0.0026/GB retrieval fee.

OCI's storage pricing is competitive, particularly when combined with the 10 TB per month of free outbound data transfer. For workloads that frequently retrieve data from object storage and serve it to external consumers (content delivery, data distribution, API responses with large payloads), OCI's combination of competitive storage pricing and free egress can provide the lowest total cost of ownership.

OCI supports object lifecycle policies that automatically transition objects between tiers and delete objects after a specified age. Policies can be applied at the bucket level with prefix-based filtering to apply different rules to different object categories within the same bucket.

Estimate OCI Object Storage costsBuild OCI object lifecycle policies

Decision Framework: When to Use Each Tier

Use the following decision framework to assign data to the appropriate storage tier:

  • Hot/Standard: Data accessed multiple times per day. Application data, active datasets, frequently read media files, recent logs. This is the default for data that your application reads regularly.
  • Infrequent Access/Cool/Nearline: Data accessed a few times per month. Older logs (30-90 days), previous versions of datasets, quarterly reports, backup snapshots less than 90 days old. The retrieval fee is minimal for occasional access.
  • Cold/Coldline: Data accessed a few times per year. Annual compliance archives, old project files, superseded datasets, previous year logs. Available immediately when needed but at lower storage cost.
  • Archive/Deep Archive: Data retained for compliance but rarely or never accessed. Multi-year compliance archives, legal holds, historical backups, regulatory data. Retrieval is slow (hours) and expensive, but storage cost is minimal.

Consider the total cost including retrieval when choosing tiers. If you access data in Glacier Deep Archive monthly, the retrieval fees ($0.02/GB) negate the storage savings compared to Glacier Instant Retrieval ($0.004/GB/month with $0.03/GB retrieval). Run the math on your actual access patterns before moving data to the cheapest tier.

Implementing Lifecycle Policies

Lifecycle policies automate the transition of objects between storage tiers based on age, access patterns, or both. Without lifecycle policies, tiering is a manual process that never gets done consistently. With lifecycle policies, tiering happens automatically and continuously.

A recommended baseline lifecycle policy for most object storage buckets: transition objects to Infrequent Access after 30 days, to Archive after 90 days, and delete after the regulatory retention period (365 days, 7 years, or never, depending on your compliance requirements). For buckets with known access patterns (e.g., log files that are only queried within the first week), configure more aggressive tiering: move to IA after 7 days and to Archive after 30 days.

Lifecycle policies should also clean up incomplete multipart uploads, which are a hidden storage cost. On S3, incomplete multipart uploads accrue storage charges but are not visible in normal bucket listings. Configure a lifecycle rule to abort incomplete multipart uploads after 7 days. On Azure, incomplete block blobs with uncommitted blocks should be cleaned up similarly.

Test your lifecycle policies on a non-production bucket first. Verify that objects transition to the correct tiers, that retrieval works correctly from each tier, and that the total cost matches your expectations. Monitor lifecycle policy execution through cloud provider metrics to confirm that transitions are happening as expected.

Cross-Cloud Storage Cost Comparison

For a 100 TB dataset with the following access pattern: 10 TB accessed daily (hot), 30 TB accessed monthly (warm), 60 TB accessed annually (cold/archive):

  • AWS: 10 TB Standard ($230) + 30 TB Standard-IA ($375) + 60 TB Glacier Instant Retrieval ($240) = $845/month
  • Azure: 10 TB Hot ($180) + 30 TB Cool ($300) + 60 TB Cold ($216) = $696/month
  • GCP: 10 TB Standard ($200) + 30 TB Nearline ($300) + 60 TB Coldline ($240) = $740/month
  • OCI: 10 TB Standard ($255) + 30 TB Infrequent ($300) + 60 TB Archive ($156) = $711/month

These costs are for storage alone and do not include retrieval fees, which depend on how frequently the warm and cold data is actually accessed. Azure is the cheapest for this profile due to the Cold tier pricing. OCI is competitive overall and offers the best value when data egress is factored in.

Start with Intelligent-Tiering or Autoclass

If you are not sure about your access patterns, enable S3 Intelligent-Tiering (AWS) or Autoclass (GCP) on all buckets. These features automatically move data to cheaper tiers with no retrieval fees and no risk of inaccessibility. You can refine your tiering strategy later based on the access pattern data these tools collect.

AWS S3 Storage Classes GuideAzure Storage Account Types GuideGCP Storage Classes and Lifecycle GuideOCI Object Storage GuideCompare object storage costs across clouds

Written by Jeff Monfield

Cloud architect and founder of CloudToolStack. Building free tools and writing practical guides to help engineers navigate AWS, Azure, GCP, and OCI.

Disclaimer: This article is for informational purposes. Cloud services and pricing change frequently; always verify with official provider documentation. AWS, Azure, GCP, and OCI are trademarks of their respective owners.