Today I Learnt: TIL: AWS S3 Files and what it could unlock for CoRE Stack builders
10 views
Skip to first unread message
Amit Kumar
Apr 9, 2026, 5:49:28 AMApr 9
to core-st...@googlegroups.com
Hello everyone,
Source: https://aws.amazon.com/blogs/aws/launching-s3-files-making-s3-buckets-accessible-as-file-systems/
Today I learnt about Amazon S3 Files, announced by AWS on April 7, 2026. It allows a standard S3 bucket to be mounted and used like a file system from EC2, ECS, EKS, and even Lambda, while still keeping S3 as the underlying storage layer. AWS says it supports NFS v4.1+ operations, can serve active data at around 1 ms latency, and lets multiple compute resources share and mutate the same bucket-backed data without duplicating it.
For CoRE Stack, this is interesting because many of our workflows still move through a cloud servers in order to store, access, collaborate and serve large geospatial datasets and are important part of our infrastructure.
That function explicitly runs `aws s3 sync` after artifacts are produced. S3 Files suggests a possible future where some of these pipelines, developer notebooks, and metadata publishing flows could operate directly on a shared bucket-mounted workspace. For developers, that could reduce glue code, simplify artifact sharing across containers or VMs, and make experimentation easier.
Of course, we would still need to evaluate consistency, cost, security, and whether this fits our actual access patterns before adopting it. But as an idea, it is highly relevant for open geospatial infrastructure.
Source: https://aws.amazon.com/blogs/aws/launching-s3-files-making-s3-buckets-accessible-as-file-systems/
Today I learnt about Amazon S3 Files, announced by AWS on April 7, 2026. It allows a standard S3 bucket to be mounted and used like a file system from EC2, ECS, EKS, and even Lambda, while still keeping S3 as the underlying storage layer. AWS says it supports NFS v4.1+ operations, can serve active data at around 1 ms latency, and lets multiple compute resources share and mutate the same bucket-backed data without duplicating it.
For CoRE Stack, this is interesting because many of our workflows still move through a cloud servers in order to store, access, collaborate and serve large geospatial datasets and are important part of our infrastructure.
That function explicitly runs `aws s3 sync` after artifacts are produced. S3 Files suggests a possible future where some of these pipelines, developer notebooks, and metadata publishing flows could operate directly on a shared bucket-mounted workspace. For developers, that could reduce glue code, simplify artifact sharing across containers or VMs, and make experimentation easier.
Of course, we would still need to evaluate consistency, cost, security, and whether this fits our actual access patterns before adopting it. But as an idea, it is highly relevant for open geospatial infrastructure.
If this sparks an idea, contribute to CoRE Stack and bring it to our next Dev Community call tomorrow, Friday, April 10, 2026 at 3 PM.
- Amit
Reply all
Reply to author
Forward
0 new messages