One connection, several datasets

A BigQuery connection can now span several datasets, and a PostgreSQL connection several schemas — no more one connection per dataset, and joins across them just work.

Pick your datasets

After Test Connection, the form lists every dataset (or schema) your credentials can see. Tick the ones to analyze; the first becomes the primary, whose tables keep their plain names. On BigQuery, all ticked datasets must share a location — Axiome checks before saving.

Tables from the other datasets

Tables outside the primary get a semantic model named dataset__table — crm__customers for the customers table of crm — so same-named tables never collide, and the Semantic models page groups models by dataset. Existing connections and widgets are untouched: adding a dataset later changes nothing for what you already built.

Axi and dbt follow

Axi sees every dataset on the connection and writes its SQL accordingly. dbt imports now bind models in any of the connection's datasets — marts in one, staging in another — instead of the primary one only, and the import report names the datasets the connection covers when a model falls outside them.

📖 Managing connections · Semantic layer overview · Import from dbt