S3 Files launches, letting you use S3 buckets like a file system

Amazon S3 Files connects an S3 bucket as a file system, greatly lowering the boundary between object storage and file systems. Changes are automatically synchronized between the S3 bucket and the file system, and multiple compute resources can share the same data without duplication.

Now you can handle S3 data like ordinary files from EC2, ECS, EKS, and Lambda. Files and directories are exposed via NFS v4.1+, supporting operations such as create, read, modify, and delete.

The way it works is split into two layers. Frequently used data is stored in the file system's high performance storage to provide low latency, while cases like large sequential reads are served directly from Amazon S3 to increase throughput. For byte-range reads, only the requested portion is transferred, reducing data movement and cost.

Key features are as follows.

• ~1ms latency access to active data
• Concurrent access from multiple compute resources and NFS close-to-open consistency
• Per-file control of loading full data or metadata only
• TLS 1.3 encryption in transit
• Encryption at rest based on SSE-S3 or AWS KMS
• IAM-based file system/object-level permission control
• Monitoring and auditing via CloudWatch and CloudTrail

Getting started is also simple. In the console, press Create file system, specify the bucket name, check the automatically created mount target, and connect it to an EC2 instance with mount -t s3files. When using the CLI, file system creation and mount target creation are separate commands.

The author explains that S3 Files is especially suited to interactive workloads, agentic AI, and ML training pipelines where multiple systems read and write together. Conversely, for existing NAS migrations, HPC, or cases requiring specific file system features, they note that the Amazon FSx family is a better fit.

Currently, S3 Files is available in all commercial AWS Regions, and billing applies to the amount of data stored in the file system, small file reads and all write operations, and S3 requests that occur during synchronization.