Multimodal data
Quickly ingest any data, any size.
Maximize
Model FLOPs
Utilization
Multimodal data
Quickly ingest
any data,
any size.
Process and Enrich
Append columns
(& rows) without
rewriting existing data
Process and Enrich
Enrich objects with
properties
Scale without worry
Scale to millions of
columns, without
upfront schema design
Avoid I/O bottlenecks
Saturate your GPUs
Run an interactive query that loads more bytes per second into an H100 than if you precompute the result and save it as Parquet on local disk.
No more custom
data access layers
All the flexibility you need, out of the box.
Selective read
Parameterized push-down read
Integrates with tools you love
Familiar, interoperable standards.
FOR COMPLEX DATA
AT MACHINE SCALE
Spiral is built on Vortex
Based on decades of database research, Vortex is the next-generation, open-source columnar format. We designed it to bridge storage & compute for the next generation of data processing, and were honored to donate it to the Linux Foundation to maximize its impact.
- Pareto-optimal performance: Faster than Apache Parquet for virtually any workload.
- Future-proof: Intentionally extensible to stay on the bleeding edge.
- Interoperable: Works seamlessly with existing data ecosystems and tools.