Amazon Web Services today launched Amazon S3 Files, a groundbreaking service that turns S3 buckets into fully functional, high-performance file systems accessible from any AWS compute resource. This eliminates the long-standing dilemma between object storage economics and file system interactivity.
Breaking the Object vs. File Barrier
“Until now, customers had to choose between S3’s cost and durability or a file system’s interactive capabilities,” said an AWS spokesperson. “S3 Files removes that tradeoff, making S3 the central hub for all data—directly accessible from EC2, ECS, EKS, or Lambda.”
Background on this shift explains why it matters.
The service supports all NFS v4.1+ operations, allowing users to create, read, update, and delete files as if they were on a local drive. Changes are automatically synced back to the S3 bucket, enabling fine-grained control over data.
How It Works
S3 Files uses a high-performance local cache to store frequently accessed file metadata and data. Files needing low-latency access are served from this cache, while large sequential reads are streamed directly from S3 at maximum throughput. Byte-range reads only transfer requested bytes, minimizing costs.
“We’ve built intelligent pre-fetching and fine-grained control over what lands on the high-performance storage,” the spokesperson added. “You can load full file data or just metadata, optimizing for your specific access patterns.”
Under the hood, the system leverages Amazon Elastic File System (EFS) for its architecture, ensuring seamless scaling and durability.
Background
For over a decade, object storage and file systems were fundamentally different. An AWS trainer once compared S3 objects to library books—you must replace the whole book to edit a page—while files on a computer are editable page-by-page. This drove customers to choose separate storage for analytics, machine learning, and production workloads.
S3 Files blurs that line, allowing any general-purpose bucket to mount as a native file system across multiple compute instances, containers, or functions. Data sharing across clusters no longer requires duplication.
What This Means
Enterprises can now unify data lakes, AI training pipelines, and real-time applications under a single S3 namespace. Machine learning teams can iterate on models without migrating preprocessed data, while DevOps runs containerized apps with consistent file access.
“This is a paradigm shift—S3 becomes the control plane for all data, regardless of access pattern,” the AWS spokesperson said. “Customers can reduce storage complexity while maintaining the performance they expect from a file system.”
Analysts predict widespread adoption, especially for hybrid cloud deployments and edge computing where low-latency file access over S3’s global infrastructure is critical.
For more details, see the official announcement.