Product architecture

Efficient execution of ad-hoc queries requires sophisticated query planning and rewriting algorithms, which in turn depend on accurate statistics, extended metadata and AI inference. Iceberg and Delta Lake are primitive and missing this metadata.

Floecat’s an open source meta-catalogue for Iceberg and Delta Lake that transparently enriches your lakehouse with the missing statistics and metadata. It automatically keeps statistics up to date, even with petabyte-scale data sets. Floecat GitHub.

Running ad-hoc SQL directly on your lakehouse needs a powerful, scalable compute engine with a real database feel.

FloeSQL has solved the difficult metadata scalability issues with lakehouses. It uses intelligent caching and LLVM-based vectorized execution to deliver the query execution speed your business users expect. 

With its powerful query planner, FloeSQL executes queries with lots of joins and complicated SQL syntax without breaking your budget.

Workload management lets you assure quality of service by isolating workloads, penalty-boxing bad queries and controlling compute resources.

Floe compute resources scale horizontally and without interruption, from one user with a small data set to thousands of concurrent users, tens of thousands of tables and petabytes of data.

Floe natively talks PostgreSQL dialect, and supports *DBC, Arrow, MCP and Python client connectivity. From Claude and ChatGPT to Metabase, Microstrategy, PowerBI and Tableau, we have you covered.