AWS lets S3 behave like a filesystem
AWS announced Amazon S3 Files, which allows S3 buckets to be mounted as native file systems with sub‑millisecond access and POSIX compatibility — the biggest change to S3 in about 20 years. Making object storage look and feel like a local filesystem could simplify migrations, reduce infrastructure layers and change how applications access cloud data. (x.com) (x.com)
For 20 years, Amazon Simple Storage Service was the place you threw files when you wanted cheap, durable storage, not the place you pointed old file-hungry software. On April 7, 2026, Amazon Web Services said you can now mount that storage as a real file system called Amazon S3 Files. (aws.amazon.com) Object storage and file storage have always spoken different languages. Amazon Simple Storage Service stores blobs with names like a warehouse shelf label, while a file system expects folders, file locks, and commands like “open,” “rename,” and “append.” (aws.amazon.com) That mismatch is why companies kept building extra layers. They would keep the master copy in Amazon Simple Storage Service, then add Amazon Elastic File System, Amazon FSx for Lustre, caches, or custom sync jobs so applications could see the same data as files. (aws.amazon.com 1) (aws.amazon.com 2) Amazon Web Services tried a partial fix in 2023 with Mountpoint for Amazon Simple Storage Service. That tool let Linux programs mount buckets, but Amazon said it was built for large, read-heavy workloads and did not try to emulate full Portable Operating System Interface behavior. (aws.amazon.com 1) (aws.amazon.com 2) Amazon S3 Files is the bigger step because Amazon says it gives “fully-featured” file system access with Portable Operating System Interface semantics, including reading, writing, and locking files. In plain English, older software can talk to a bucket more like it talks to a network drive. (aws.amazon.com) (docs.aws.amazon.com 1) (docs.aws.amazon.com 2) Amazon says the file system is built using Amazon Elastic File System, but the data stays in Amazon Simple Storage Service instead of being copied into a separate store. The file system keeps a fraction of active data as files and directories for low-latency access, while the bucket remains the system of record. (docs.aws.amazon.com) (docs.aws.amazon.com) The speed claim is about one millisecond latency for file access on Amazon Web Services compute, which is far faster than the “fetch an object over an application programming interface” pattern many cloud apps were written around. Amazon also says multiple compute resources can attach to the same S3 Files system and share data without duplication. (aws.amazon.com) (aws.amazon.com) The trick is synchronization. When you first touch a directory, Amazon S3 Files imports metadata for all files in that directory and, by default, pulls in data for files smaller than 128 kilobytes; larger files are read directly from Amazon Simple Storage Service unless you raise that threshold. (docs.aws.amazon.com) (docs.aws.amazon.com) That means the product is not pretending every byte lives on a local disk. It is acting more like a smart front desk that keeps the small, frequently touched items nearby and reaches back into the warehouse for the big boxes. (docs.aws.amazon.com) (docs.aws.amazon.com) Amazon Simple Storage Service turned 20 on March 14, 2026, so this launch lands almost exactly at the service’s 20-year mark. That timing helps explain why people inside Amazon are describing it as one of the biggest changes to Amazon Simple Storage Service since the product launched in 2006. (aws.amazon.com) (aws.amazon.com) The immediate winners are teams with old enterprise software, analytics pipelines, and artificial intelligence training jobs that still expect a file path instead of an object call. If those workloads can point at Amazon Simple Storage Service directly, one bucket can become both the archive and the live workspace. (aws.amazon.com) (docs.aws.amazon.com)
