Amazon S3 Turns 20 Storing Over 500 Trillion Objects Across Hundreds of Exabytes as AWS Rewrites Core Components in Rust
Amazon S3 marks its 20th anniversary storing over 500 trillion objects across hundreds of exabytes, while AWS progressively rewrites core storage components in Rust for memory safety and performance.
Amazon Web Services celebrated the 20th anniversary of Amazon Simple Storage Service on March 14, a date chosen for the original 2006 launch because it falls on Pi Day. What began as a service running on roughly 400 storage nodes across 15 racks in three data centers with approximately one petabyte of total capacity and 15 Gbps of bandwidth has grown into what AWS describes as the backbone of the modern internet.
S3 now stores more than 500 trillion objects, serves over 200 million requests per second globally, and holds hundreds of exabytes of data across 123 Availability Zones in 39 AWS Regions. The maximum object size has increased from 5 GB at launch to 50 TB today, a 10,000-fold expansion. Pricing has dropped to just over two cents per gigabyte per month, representing an approximately 85 percent reduction from the original rates.
Twenty Years of API Stability
AWS principal developer advocate Sebastien Stormacq highlighted backward compatibility as one of S3’s most significant achievements. “The code you wrote for S3 in 2006 still works today, unchanged,” Stormacq noted, pointing to the complete preservation of the original API across two decades of underlying infrastructure changes.
That stability has turned the S3 API into a de facto industry standard. Multiple cloud providers and storage vendors now offer S3-compatible interfaces, making it one of the most widely implemented APIs in cloud computing. The service’s consistency guarantees were upgraded in December 2020 when AWS introduced strong read-after-write consistency for all S3 operations at no additional cost, eliminating a long-standing limitation that had required developers to build workarounds for eventual consistency.
Rust Rewrites and Durability Engineering
Behind the scenes, AWS has been progressively rewriting S3’s performance-critical code in Rust. Blob movement and disk storage components have been rebuilt in the systems programming language, which provides memory safety guarantees without the overhead of garbage collection. The effort reflects a broader industry trend toward Rust adoption in infrastructure software, and AWS has been among the most prominent advocates of the language for production systems.
S3 maintains its 11-nines durability rating, meaning 99.999999999 percent of objects are preserved, through continuous microservices auditing of every stored byte. AWS has described the system as performing constant background verification to detect and repair any data integrity issues before they affect customers.
From Storage to Data Lake Foundation
The anniversary also underscored how S3’s role has expanded far beyond simple object storage. AWS VP and Distinguished Engineer Andy Warfield discussed at the Pi Day event how S3 now powers over a million data lakes and serves as foundational infrastructure for AI workloads globally. The Hadoop S3A connector was instrumental in this transformation, allowing the Hadoop ecosystem to use S3 as underlying storage and expanding its purpose from a file repository to what Warfield described as a shared data foundation where analytics engines, applications, and research workflows operate against the same datasets.
S3 now offers multiple storage classes optimized for different access patterns, from S3 Standard for frequently accessed data to S3 Glacier Deep Archive for long-term retention at the lowest cost. The introduction of S3 Intelligent-Tiering automatically moves objects between access tiers based on usage patterns, reducing storage costs without requiring manual lifecycle management.
Security Legacy
S3’s history has not been without friction. The service’s original default of allowing public access to buckets led to a series of high-profile data exposure incidents over the years, as misconfigured buckets exposed sensitive data belonging to governments, enterprises, and millions of individuals. AWS has since changed the default to block public access and introduced multiple layers of access controls, encryption options, and monitoring tools to prevent accidental exposure.
The 20th anniversary arrives as cloud storage demand continues to accelerate, driven in large part by AI training workloads that require massive datasets. S3’s scale, serving more than a quadrillion requests annually, positions it as infrastructure that most internet users interact with daily, whether they are aware of it or not.