Apache-2.0 · A Linux Foundation project
Fast, small, open: pick three.
Vortex is an open-source columnar file format for fast scans, random access, and compression.
$ cargo add vortex$ uv add vortex-data$ uvx --from vortex-data vx --helpINSTALL vortex; LOAD vortex; SELECT * FROM read_vortex('data.vortex');$ cargo add vortex-datafusionimplementation("dev.vortex:vortex-spark_2.13:0.90.0:all")
Vortex against Parquet, Lance, and Arrow · times in ms, ratios to Parquet, best first. A time is the geomean over the queries or datasets that bench's score covers. A ratio is against Parquet = 1.
bench.vortex.dev · readScan throughput
5.2× faster than Parquet
Write throughput
2.0× faster than Parquet
ClickBench on DataFusion
56% faster than Parquet
GPU analytics with cuDF, end to end, hot, on GH200, NDS-H SF10, Parquet the fastest of four configurations per query, from the Vortex UW–Madison talk
3.9× faster than Parquet
Random access, NVMe, hot
103× faster than Parquet
Size vs Parquet+Zstd
14% larger than Parquet
