Dataset lineage
On a dataset's Lineage tab, review the pipelines that read or write the dataset and its upstream and downstream assets.
Review lineage
- Select the dataset in the collection tree.
- Open the Lineage tab.
- Check the number of upstream and downstream assets and the number of pipelines.
- Select the traversal depth: 1 hop, 2 hops, or All. The default is 2 hops.
- Select an asset or pipeline in the graph to go to its screen.
Pipelines that write this dataset as an output, and their input assets, appear on the left as upstream. Pipelines that read this dataset as an input, and their output assets, appear on the right as downstream.
If there are connections beyond the current depth, a truncation notice appears below the graph. To see further, select All.
Display scope
Lineage is built from the inputs and outputs of pipelines the current user can read. The downstream count shown is therefore the minimum that can be confirmed.
The following connections do not appear in the graph.
- Pipelines you do not have permission to read
- Transformations inside code steps
- In-memory
variablechannels - Upstream sources beyond a connector
- Consumption by dashboards or agents
An empty graph is not proof that no pipeline references this dataset. Check both the pipeline connections and the user's read permissions. A pipeline outside those permissions may exist, or the dataset may be used in a way that lineage does not display.
Difference from the Used By tab
| Tab | What it shows |
|---|---|
| Used By | Ontology entities and relationships that use this dataset as a source. |
| Lineage | Datasets, sources, views, and pipelines connected through pipeline inputs and outputs. |
Check both tabs before deleting a dataset or changing its structure.
If an error occurs
If Failed to load lineage appears, select Retry. If it keeps failing, send the dataset ID and the error message to your administrator.