Explore Vortex
Vortex is a highly performant, extensible columnar file format, with an associated toolkit for running compute over compressed data in-memory.
Your data without limits
Vortex is the fastest open-source columnar format, designed from the ground up to achieve optimal performance on modern hardware. Now a sub-project of the Linux Foundation.
Pareto-optimal Performance
- 100x faster random access.
- 10-20x faster scans.
- 5x faster writes.
- Comparable size to modern Parquet.
Cascaded Lightweight Encodings
- "White box" encodings support pushdown compute on compressed data.
- Designed for random access & decompression via SIMD (CPU) & SIMT (GPU).
- Based on the latest research, like FastLanes, ALP, FSST, & BtrBlocks.
Extensible & Future-proof
- Rich extension types.
- Pluggable encodings & compression strategies.
- Self-describing layouts & WASM decoders allow the format to evolve.
Optimized Metadata
- Designed for efficient reads from object storage.
- Zero-copy deserialization efficiently supports wide schemas.
Interoperable
- Decompressed arrays are zero-copy to/from Apache Arrow.
- Integrates with Polars, Pandas, DataFusion, DuckDB, Spark, and many other popular frameworks.