Documentation

faucet-stream is a data-movement platform. One faucet binary moves data between 43 sources and 35 sinks from a YAML file, with incremental state, CDC and exactly-once delivery built in.

Get startedLearn the architecture

cargo install faucet-cli
faucet init my_pipeline --source postgres --sink bigquery
faucet run my_pipeline.yaml

What do you want to do?

Move your first dataInstall faucet and run a pipeline in five minutes. Keep tables in syncIncremental replication, upserts and CDC mirrors. Run it in productionDeploy, schedule, monitor and upgrade safely. Find a connectorBrowse every source and sink and what each supports.

• Shape data with transforms
• Check data quality
• Use Template Hub templates
• Embed faucet in Rust
• Write your own connector
• Compare with other tools

Understand the foundations

Four ideas explain almost everything faucet does.

How a pipeline runsSource, transforms and sink, streamed page by page in bounded memory. State and resumingBookmarks that let every run pick up exactly where the last one stopped. Delivery guaranteesAt-least-once by default, exactly-once where the sink can commit atomically. Where your data may goData-flow policies, PII masking and contracts, checked before a row moves.

Where else to look

API docsEvery crate on docs.rs, with all features on. Template HubReady-made source and sink templates. Runnable examplesA config for nearly every connector pair.