F3: A Future-Proof Open-Source Data File Format
F3 is an open-source data file format, presented as a SIGMOD 2026 paper, aimed at next-generation storage workloads as an evolution beyond Parquet. It is released under the MIT License and hosted on GitHub at future-file-format/F3.
Core Design Principles
- Efficiency: Vectorized decoding, with read throughput outperforming Parquet.
- Interoperability: Self-describing files with native cross-platform compatibility.
- Extensibility: A generic API supporting custom encodings.
- Read throughput: better than Parquet
- Random-access latency: significantly reduced
- Metadata parsing: over 10x faster
Key Innovation: Embedded Decoders
The standout technique is WebAssembly embedding: each file carries its own decoding logic alongside data and metadata (a DATA / META / WASM layering). This means even older readers can decode newer encoding formats, addressing format evolution and version compatibility at only a KB-scale storage overhead.
Architecture: IOUnit Decoupled Design
F3 decouples storage into IOUnits, solving Parquet's fixed row-group size limitation.
Reported Performance Advantages
Summary
By combining vectorized decoding, self-describing interoperable files, extensible custom encodings, and embedded WASM decoders for backward compatibility, F3 positions itself as a future-proof successor to Parquet for modern data analytics pipelines.