Amazon S3 Weekly — 2026-04, Week 15
Editor’s Note
This week brings meaningful shifts in how practitioners interact with S3, from native NFS 4.1 filesystem semantics to interesting approaches for versioned state management and bandwidth-efficient synchronization. The convergence of object storage with traditional filesystem protocols continues to reshape architectural patterns for cloud-native workloads.
Top Stories
S3 Buckets Now Mount as NFS 4.1 Filesystems
AWS has introduced the capability to mount any S3 bucket as an NFS filesystem across EC2, ECS, EKS, and Lambda environments. The implementation adheres to the NFS 4.1 protocol standard, allowing existing tools and workflows built around POSIX filesystem semantics to operate directly against object storage without modification. This changes the access model from object-only APIs to traditional filesystem operations, potentially simplifying migration paths for applications that expect hierarchical directory structures and file-based I/O. The architectural implications center on workload portability and the ability to leverage S3’s durability characteristics while maintaining familiar filesystem interfaces (read more).
T4 Experimental Datastore Uses S3 as Write-Ahead Log
An experimental datastore named T4 writes its append-only write-ahead log segments and snapshots directly to S3 rather than local disk, treating object storage as the primary durable medium. This design enables native versioning, copy-on-write branching, and point-in-time state reconstruction from immutable checkpoints stored in S3. The project originated as a potential replacement for etcd in Kubernetes control planes, where immutable versioned state and low operational overhead are valued. By avoiding local disk entirely and relying on S3’s consistency guarantees, T4 explores an architecture where storage durability and versioning become properties of the underlying object store rather than application-managed features (read more).
Differential Sync Protocol Reduces S3 Transfer by 95%
Community benchmarks of the rac-delta differential synchronization protocol demonstrate significant bandwidth savings when syncing large directories with incremental changes to S3. Testing with a 2.2 GB directory containing minor modifications showed a 94.7% reduction in download transfer volume, from 2,219 MB using full-file transfers to 116 MB with content-addressed chunking based on Blake3. The corresponding egress cost savings measured 19× lower per thousand users, dropping from €184.27 to €9.66. The protocol’s chunking strategy calculates block-level deltas rather than replacing entire files, making it suited for workloads where large datasets change incrementally and egress costs dominate total cost of ownership (read more).
Worth Reading
- S3 Files and the Changing Face of S3 — Werner Vogels discusses architectural shifts in S3 access patterns
- bucket-delta-cli — Command-line tool for differential bucket synchronization
- Locker — S3-based storage utility