LF AI and Data Hosts Vortex Project for Data Access for AI and Analytics
18 days ago
- #AI
- #Data Storage
- #Open Source
- The LF AI & Data Foundation announced the launch of the Vortex Project, an open, extensible columnar storage format designed for high-performance data systems.
- Vortex is contributed by SpiralDB and supported by industry leaders like Microsoft, Snowflake, Palantir, and NVIDIA.
- Vortex offers state-of-the-art performance, including 100x faster random access reads and 5x faster writes compared to Apache Parquet.
- The format is optimized for modern workloads, including GPU-based training and cloud object storage like S3 and GCS.
- Vortex features an extensible architecture to incorporate new compression techniques and integrates with tools like Apache Arrow and Apache Spark.
- The project aims to eliminate CPU bottlenecks by enabling direct GPU decompression, improving AI training data access.
- Industry leaders highlight Vortex's potential to address storage bottlenecks and advance open-source data processing for AI.