For assets that produce database tables, column-level lineage can be a powerful tool for improving collaboration and debugging issues. Column lineage enables data and analytics engineers alike to understand how a column is created and used in your data platform.
Emitted as materialization metadata, column lineage can be:
Specified on assets defined in Dagster
Enabled for assets loaded from integrations like dbt
Dagster uses this metadata to display the column's upstream and downstream dependencies, accessible via the asset's details page in the Dagster UI. Note: Viewing column-level lineage in the UI is a Dagster+ feature.
For each column, use TableColumnDep to define its dependencies. This object accepts asset_key and column_name arguments, allow you to specify the name of the asset and column that make up the dependency.