Overview
- AWS introduced S3 Files on Tuesday, making S3 buckets behave like NFS v4.1 file systems for EC2, containers, and Lambda, with general availability in 34 regions.
- Built on Amazon EFS, the service caches hot data for low latency and streams large reads from S3, while S3 stays the source of truth through staged, whole-object commits.
- Independent measurements found a fixed 60-second push for file writes to reach S3, about 30 seconds for new S3 objects to show up on mounts, and roughly two seconds for updates to cached files.
- Tests also surfaced pitfalls such as certain object key names becoming invisible in the mounted view, deleted objects remaining readable for 6 or 18 seconds, and write access failing when access-point UIDs conflict with imported file ownership.
- AWS points to CloudWatch metrics for import failures and plans richer logs and API-level sync controls, as security researchers urge tighter VPC rules and IAM alignment to curb mount exposure and ransomware-style risks.