Amazon S3
Amazon Simple Storage Service (S3), launched in 2006 and headquartered in Seattle, Washington, is the foundational object storage platform within Amazon Web Services.
Publisher review
Amazon Simple Storage Service (S3), launched in 2006 and headquartered in Seattle, Washington, is the foundational object storage platform within Amazon Web Services. It is designed for developers, data engineers, and enterprises that need to store and retrieve any amount of data from anywhere — from website content and backup archives to big-data analytics lakes and machine learning datasets. S3 is the default choice for many AWS-native workloads because it integrates directly with virtually every other AWS service (Lambda, EMR, Redshift, etc.) and offers eight distinct storage classes to match access patterns, from hot data to deep archive. It is also widely used as a data source for external applications via its REST API, making it a de facto standard for cloud object storage across the industry.
S3 delivers strong read-after-write consistency for all operations (added in December 2020) and is engineered for 99.999999999% annual durability, storing data redundantly across a minimum of three Availability Zones in the S3 Standard class. Performance scales automatically: each prefix supports at least 3,500 PUT/POST/DELETE and 5,500 GET requests per second, and by spreading keys across many prefixes, real-world deployments routinely handle tens of thousands of requests per second. The service also offers features like S3 Intelligent-Tiering, which uses machine learning to move objects between access tiers and optimize cost, and lifecycle policies that automate transitions to lower-cost classes or deletion. S3 Transfer Acceleration can speed up uploads over long distances, and cross-region replication adds cost for geo-redundancy.
In the cloud object-storage market, S3 competes directly with Google Cloud Storage (GCS) and Microsoft Azure Blob Storage. S3 generally exhibits very low first-byte latency (tens of milliseconds in-region) and has the most mature request-scaling architecture, though it may return temporary slowdown errors during very sudden traffic spikes. GCS offers excellent throughput for large sequential reads (capable of saturating 10 Gbps+ on a single TCP stream) but has a ~1,000 writes/second limit to the same object name. Azure Blob Storage has a soft limit of 20,000 requests per second per storage account and requires multiple accounts for higher scale. All three platforms now provide strong consistency, but S3’s ecosystem breadth and eight-class tiering give it a pricing flexibility advantage for mixed workloads.
The honest trade-offs: S3 billing is notoriously complex, comprising separate charges for storage (e.g., $0.023/GB/month in US East), requests (Class A at $5/million, Class B at $0.40/million), data retrievals, and egress fees ($0.09/GB from US East). Users report that the AWS Management Console can feel cluttered and unfriendly for non-power-users, and that speed can be inconsistent under unpredictable access patterns. While S3 is highly durable, it is not immune to common cloud problems like unexpected cost spikes from runaway requests or data transfer. For workloads with predictable access, S3 Standard costs $42.40/month for 1.8TB, but archival classes like Glacier Deep Archive drop storage cost below $1/TB/month — at the expense of retrieval times measured in hours.
How it works
-
Strong read-after-write consistency
Since December 2020, all S3 operations provide strong consistency, so reads immediately reflect writes — simplifying application logic.
-
Automatic request scaling
Supports at least 3,500 PUT/POST/DELETE and 5,500 GET requests per second per prefix, scaling automatically with workload.
-
Eight storage classes
Offers eight classes including Standard, Intelligent-Tiering, Glacier, and Deep Archive, each optimized for different access frequency and cost.
-
99.999999999% durability
Engineered for 11 nines of annual durability, storing data across a minimum of three Availability Zones in S3 Standard.
-
S3 Intelligent-Tiering
Uses machine learning to automatically move objects between access tiers, optimizing costs without manual lifecycle policies.
-
Lifecycle policies
Automates transitions between storage classes or deletion based on age or other rules, reducing manual management.
-
S3 Transfer Acceleration
Speeds up uploads and downloads over long distances using AWS edge locations, with additional cost per GB transferred.
Strengths and trade-offs
Strengths
- S3 offers 99.999999999% annual durability, meaning if you store 10 million objects, you might expect to lose one object every 10,000 years.
- Each prefix supports at least 3,500 PUT/POST/DELETE and 5,500 GET requests per second, and spreading keys across many prefixes enables tens of thousands of requests per second.
- S3 Intelligent-Tiering uses machine learning to automatically optimize costs by moving objects between access tiers without manual intervention.
- Strong read-after-write consistency for all operations (added December 2020) eliminates the need for application-level consistency workarounds.
Trade-offs
- Billing is complex, with separate charges for storage ($0.023/GB/month), requests (Class A at $5/million), and egress ($0.09/GB from US East), making cost prediction difficult.
- The AWS Management Console for S3 is often described as cluttered and unfriendly for non-power-users, with a steep learning curve for advanced features.
- During very sudden traffic spikes, S3 may return temporary slowdown errors, requiring retry logic in applications.
- Egress fees ($0.09/GB from US East) can dominate costs for data-heavy workflows, especially when moving large datasets out of AWS.
Pricing context
Storage starts at $0.023/GB/month (S3 Standard, US East); Class A requests (PUT/POST/DELETE) cost $5/million; Class B requests (GET/HEAD) cost $0.40/million; egress to internet is $0.09/GB from US East. Eight storage classes range from Standard to Glacier Deep Archive, with lower storage costs but higher retrieval fees.
Getting started with Amazon S3
-
Sign up for AWS
Create an AWS account at aws.amazon.com. Provide your email, payment information, and verify your identity. Once activated, you gain access to the AWS Management Console, where you can manage all services including S3.
-
Create an S3 bucket
In the AWS Management Console, navigate to S3 and click 'Create bucket'. Choose a globally unique name, select an AWS Region, and configure options like versioning or encryption. Click 'Create bucket' to finalize.
-
Set bucket permissions
After creating the bucket, go to the 'Permissions' tab. Define a bucket policy or configure Access Control Lists (ACLs) to control read/write access. For public access, uncheck 'Block all public access' and add a policy granting public read.
-
Upload your first object
In the bucket, click 'Upload'. Select a file from your local machine, set storage class (e.g., S3 Standard), and configure optional properties like encryption or metadata. Click 'Upload' to store the object in S3.
-
Schedule a lifecycle policy
In the bucket's 'Management' tab, create a lifecycle rule. Define transitions, such as moving objects to S3 Glacier after 30 days, and set expiration to delete objects after 365 days. Apply the rule to automate cost optimization.
Frequently Asked Questions
What is Amazon S3 and what is it used for?
Amazon S3 is a cloud object storage service launched in 2006. It stores and retrieves any amount of data from anywhere, serving use cases like website content, backup archives, big-data analytics lakes, and machine learning datasets.
What storage classes does Amazon S3 offer?
Amazon S3 offers eight storage classes, including Standard, Intelligent-Tiering, Glacier, and Deep Archive. Each class is optimized for different access frequencies and costs, from hot data to deep archive with retrieval times in hours.
How does Amazon S3 pricing work and what are the main costs?
S3 pricing includes storage at $0.023/GB/month, Class A requests at $5/million, Class B requests at $0.40/million, and egress fees of $0.09/GB. Billing is complex, with separate charges that can lead to unexpected cost spikes.
How does Amazon S3 compare to Google Cloud Storage and Azure Blob Storage?
S3 competes with GCS and Azure Blob Storage. S3 offers the most mature request-scaling architecture and eight storage classes. GCS excels in large sequential reads, while Azure has a soft limit of 20,000 requests per second per account.
What is the durability and consistency of Amazon S3?
S3 provides 99.999999999% annual durability, storing data across three Availability Zones. Since December 2020, it offers strong read-after-write consistency, meaning reads immediately reflect writes, simplifying application logic.
What are the weaknesses of Amazon S3?
S3 billing is complex with separate charges for storage, requests, and egress, making cost prediction difficult. The console is cluttered for non-power-users, and sudden traffic spikes may cause temporary slowdown errors requiring retry logic.
Alternatives in this category
How Amazon S3 compares
Direct head-to-head against 3 competitors. Picked by 7wData.
Amazon S3
- Pricing
- Storage starts at $0.023/GB/month (S3 Standard, US East); Class A requests (PUT/POST/DELETE) cost $5/million; Class B requests (GET/HEAD) cost $0.40/million; egress to internet is $0.09/GB from US East. Eight storage classes range from Standard to Glacier Deep Archive, with lower storage costs but higher retrieval fees.
- Target
- Amazon Simple Storage Service (S3), launched in 2006 and headquartered in Seattle, Washington, is the foundational object storage platform within Amazon Web Services.
- Strength
- S3 offers 99.999999999% annual durability, meaning if you store 10 million objects, you might expect to lose one object every 10,000 years.
- Watch for
- Billing is complex, with separate charges for storage ($0.023/GB/month), requests (Class A at $5/million), and egress ($0.09/GB from US East), making cost prediction difficult.
Backblaze B2
- Pricing
- $6/TB/month storage; $0.01/GB egress after 3x free; $0.004/1K writes
- Target
- Cost-sensitive teams, AI/ML workloads needing fast dataset loading
- Deployment
- Public cloud (US, NL, DE)
- Strength
- Lowest per-TB storage cost among major S3 alternatives at $6/TB/month
- Watch for
- Egress fees apply after 3x data volume free tier; limited region count
Cloudflare R2
- Pricing
- $0.015/GB/month storage; $0 egress; no request fees for first 10M/month
- Target
- Globally distributed apps needing zero egress fees and edge performance
- Deployment
- Global edge network (250+ locations)
- Strength
- Zero egress fees with sub-10ms latency from 250+ edge locations
- Watch for
- No archive tier; limited to Cloudflare ecosystem for advanced features
Wasabi Hot Cloud Storage
- Pricing
- $6.99/TB/month flat; $0 egress; $0 request fees
- Target
- Backup, archive, and media serving with predictable flat-rate pricing
- Deployment
- Public cloud (multiple regions)
- Strength
- Flat-rate pricing with no egress or API request fees
- Watch for
- 90-day minimum storage duration; no cold/archive tier
User reviews
No user reviews yet. Be the first to write one.
Sources
Reporting on this tool draws on these publicly available sources.