A data pipeline runtime for DuckDB and DuckLake.

Ingestion, transformation, validation, and scheduling in a single binary.

curl -fsSL https://ondatra.sh/install.sh | sh

terminal
Install and create a project

Write SELECT statements. OndatraSQL handles materialization, change detection, schema evolution, and dependency ordering automatically.

Fetch data from REST APIs using lib functions with built-in HTTP, OAuth, and pagination. Your SQL SELECT controls which columns are fetched.

30 constraint macros block bad data before insert. 18 audit macros catch regressions inside the transaction. 14 warning macros monitor trends.

Only changed data is processed. OndatraSQL uses DuckLake's table_changes() to detect which source rows changed between snapshots.

Run the full pipeline against a temporary catalog copy. See row diffs, schema changes, and downstream impact before committing.

One command installs an OS-native scheduler. Pipelines run on cron via systemd or launchd.

Column-level lineage

OndatraSQL traces every column from source to output.

raw.countries

name

capital

population

↓

staging.countries

name

capital

population

size CASE

↓

mart.population

size

countries COUNT

total_pop SUM

One binary. Define pipelines in SQL. Run them anywhere.