← Back to Glossary

Rill

Rill is an open-source BI tool built on DuckDB (and optionally ClickHouse) that turns SQL models and YAML metrics definitions into fast, interactive dashboards.

Overview

Rill (from Rill Data) is a business intelligence tool aimed at giving analysts sub-second, exploratory dashboards without standing up a separate cloud warehouse for every project. It's built "BI as code": data sources, SQL transformation models, and metrics definitions live as files in a project directory, so they can be version-controlled and code-reviewed like any other engineering artifact, then compiled into interactive dashboards.

Why DuckDB

Rill Developer (the open-source, local-first product) uses DuckDB as its default embedded OLAP engine. When Rill's creators evaluated engines for a tool that needed to feel instant while iterating on a dashboard, DuckDB's single-node analytical query performance made it the natural fit — data gets ingested from sources (files, object storage, warehouses) into DuckDB, and every filter or drill-down in the dashboard becomes a DuckDB query running locally. For larger deployments, Rill also supports ClickHouse as an alternative OLAP engine, effectively letting a project scale from a laptop prototype to a much larger production dataset without rewriting the metrics layer.

Copy code

# rill metrics view definition (simplified) type: metrics_view model: orders_model dimensions: - column: country measures: - name: total_revenue expression: SUM(order_total)

Copy code

-- A Rill model, just a DuckDB SQL query SELECT order_id, country, order_total FROM read_parquet('s3://bucket/orders/*.parquet')

Where it fits

Rill sits closer to the data than a traditional dashboarding tool that only queries an existing warehouse — it does its own ingestion and transformation, then exposes a metrics layer that both dashboards and, increasingly, AI agents can query consistently. For datasets under roughly 50GB, keeping data in DuckDB is Rill's recommended default; beyond that, teams typically move to the ClickHouse engine option.

Related terms

FAQS

Yes, Rill Developer is open source. Rill Data also offers a managed cloud product for deploying and sharing Rill dashboards.

Rill's team chose DuckDB for its single-node analytical query speed, which lets dashboards feel instant while iterating locally on smaller-to-medium datasets, typically under about 50GB.

Yes. Rill also supports ClickHouse as an alternative OLAP engine for larger-scale or production deployments, while keeping the same metrics-as-code project structure.