High-performance column store
Pixels is optimized for multiple storage systems including S3, HDFS, POSIX file systems, NVMe SSDs, and other cloud-native or on-premises environments.
Explore the main repoOpen analytics platform for data lakes and warehouses
PixelsDB
High-performance storage, serverless elasticity, freshness, and interactivity.
The PixelsDB project is centered on Pixels, a columnar storage and compute engine for analytical workloads. It is designed for on-premises and cloud-native storage, integrates with mainstream query engines, and extends into serverless query processing, web-based exploration, and CDC-driven data freshness.
Overview
Main features of PixelsDB
PixelsDB combines the core Pixels engine with elasticity, transactional freshness, and interactive query experiences for modern analytical systems.
Turbo
Turbo provides serverless elastic query processing, combining autoscaling execution with serverless acceleration for bursty analytical workloads.
Read the Turbo moduleRetina
Retina provides freshness and ACID transactions, enabling real-time synchronization and consistent analytical access on evolving data.
See Retina in PixelsRover
Rover provides flexible SLAs and interactivity through a web-based experience for query execution, exploration, and natural-language-assisted workflows.
Open Pixels-RoverWhy PixelsDB
Built for modern analytical systems
Lakehouse-friendly storage
Pixels is designed for analytical tables in data lakes and warehouses, with a focus on practical deployment across cloud-native and on-premises environments.
Performance-oriented format
The project emphasizes high-throughput analytical access and reports up to two orders of magnitude improvement over Parquet in targeted scenarios.
Open ecosystem integrations
Connectors and related projects link PixelsDB to engines and tooling already common in modern data platforms.
Research-backed roadmap
The system evolves alongside published work on storage layout, serverless analytics, and freshness for lake-style data systems.
Research
Publications behind the project
PixelsDB draws from a series of research efforts in data lakes, wide-table layout optimization, and serverless analytics.
Browse the full PixelsDB research page
See all papers, demos, abstracts, and recent work on the dedicated research page.