Lineage

Column lineage

Understand column evidence, precision, and static enrichment.

Embrasure combines column mappings from these evidence paths:

  1. SQL parsing for supported warehouse query history.
  2. Warehouse-native lineage from Snowflake and Databricks.
  3. The OpenLineage columnLineage facet from runtime producers.
  4. Column links from the Google Cloud Data Lineage API through a BigQuery connector.
  5. Compiled SQL and declarations from dbt and SQLMesh artifacts.
  6. Enabled Fivetran mappings.
  7. Semantic-model expressions from supported BI systems.

Read precision

StateAction
resolvedUse the mapping for direct column impact.
partialInspect unresolved inputs and outputs.
starReplace wildcards when exact expansion matters.
unparseableCheck the SQL dialect or provider payload.
truncatedNarrow the request or page through the remainder.

Scope impact by repeating column_ref:

curl --get https://api.embrasure.ai/v1/lineage/impact \
  --header "Authorization: Bearer $EMBRASURE_API_TOKEN" \
  --data-urlencode "workspace_id=$WORKSPACE_ID" \
  --data-urlencode "connector_id=$CONNECTOR_ID" \
  --data-urlencode "relation_ref=catalog.schema.orders" \
  --data-urlencode "column_ref=customer_id" \
  --data-urlencode "column_ref=total_amount" \
  --data-urlencode "direction=downstream"

Change review reports exact column impact only when the graph has confirmed column evidence. Table-only evidence stays labeled at table precision.

Spark and Dataproc lineage

Emit OpenLineage from Spark jobs that read one catalog and write another.

Lineage policies

Choose which lineage evidence each connector shows, without deleting anything.

Read precision