Lakehouse

One lakehouse, your choice of engine: Spark, Python, and DuckDB join Trino

The Databasin lakehouse is now truly multi-engine. Run Trino for federated SQL, Spark and Python for heavy processing and notebooks, or DuckDB for fast lightweight analytics — all over the same open Iceberg tables, switchable from a single selector.

A lakehouse should let you pick the right tool for the job. As of this winter, Databasin does — across three engines and counting, all reading the same open tables.

The lineup

  • Trino — the federated SQL workhorse you already know
  • Apache Spark — clusters with notebooks and one-click wake, for ML and heavy processing (shipped late January)
  • Python — first-class support in the lakehouse for data science workflows
  • DuckDB — new today, with an in-editor engine selector: blazing, lightweight analytics with nothing to manage

One lake underneath

Every engine queries the same data — no copies, no per-engine silos, no sync jobs. New clusters run on Apache Iceberg, the open table format, so nothing about your storage is proprietary.

The supporting cast matters too: notebook automation (February 11) lets notebooks run on schedules, notebook sharing (February 15) makes them collaborative, and the multi-engine onboarding flow lets new organizations pick their engine mix at setup.

Update, June 2026: the lineup grew again — Apache Doris joined as engine number four.

← Bring any API: OpenAPI import for custom pipelines All announcements SQL results that stream: see rows the moment they exist →

See it on your own data — five minutes, $50 in credit, no card.