Impact Analysis & Dependencies

Why impact analysis matters

The lineage view

The lineage view is a visual graph showing how items connect:

Lakehouse → Dataflow Gen2 → Warehouse → Semantic Model → Report
                                              ↓
                                          Dashboard

What the lineage view shows

• Upstream sources — where data comes from
• Downstream consumers — who depends on this item
• Cross-workspace dependencies — items in other workspaces that reference this one
• External sources — connections to data outside Fabric

Accessing lineage view

1. Open the workspace
2. Click Lineage view in the top menu
3. The graph shows all items and their connections

Impact analysis for each item type

The exam tests impact analysis across four item types:

Changed Item	Downstream Affected	Example Impact
Lakehouse	Dataflows, notebooks, warehouse views, semantic models (via SQL endpoint), reports	Renaming a table breaks all SQL queries referencing it
Warehouse	Semantic models (via SQL/default model), reports, downstream warehouses (cross-DB)	Changing a view breaks the semantic model
Dataflow Gen2	Lakehouses, warehouses (where data is loaded)	Changing output schema breaks destination tables
Semantic model	Reports, dashboards, other models (via DirectQuery chaining)	Renaming a measure breaks all reports using it

💡 Scenario: Anita renames a lakehouse table

Anita at FreshCart plans to rename sales_raw to bronze_pos_transactions in her lakehouse. Before making the change, she opens the lineage view and sees:

• 2 Spark notebooks read from sales_raw
• 1 Dataflow Gen2 loads data into sales_raw
• The lakehouse SQL endpoint exposes sales_raw as a view
• A semantic model in Direct Lake mode references the SQL endpoint view
• 5 reports connect to that semantic model

Impact: renaming the table would break the notebooks, dataflow, and SQL endpoint view — which cascades to the semantic model and all 5 reports. She coordinates the change with her team before proceeding.

Best practices for managing dependencies

1. Check lineage before any schema change — column renames, table drops, type changes
2. Use views as abstraction layers — if reports connect to views (not tables), you can change tables without breaking reports
3. Notify downstream owners — the impact analysis panel shows item owners and their contact info
4. Test in Dev first — use deployment pipelines to validate changes before promoting to Prod
5. Version your changes — Git integration tracks what changed and when, enabling rollback

Question

What is the lineage view?

Click or press Enter to reveal answer

Answer

A visual dependency graph showing how Fabric items connect — upstream sources, downstream consumers, and cross-workspace dependencies. Access it from the workspace menu.

Click to flip back

Question

What does impact analysis answer?

Click or press Enter to reveal answer

Answer

'If I change X, what breaks?' It traces the dependency chain from a source item through intermediate items to end consumers (reports, dashboards). Check it before any schema or naming change.

Click to flip back

Knowledge Check

Anita plans to rename a column in a lakehouse table. What should she do FIRST?

ARename the column and test if reports still workBUse the lineage view to identify all downstream items that reference this columnCCreate a backup of the lakehouseDAsk the Fabric admin to approve the change

Check Answer

Question

Which four item types does the DP-600 exam test for impact analysis?

Click or press Enter to reveal answer

Answer

Lakehouses, warehouses, Dataflow Gen2, and semantic models. Each has different downstream dependants: lakehouses affect notebooks, dataflows, and SQL endpoints; warehouses affect semantic models and reports; dataflows affect destination tables; semantic models affect reports and dashboards.

Click to flip back

Knowledge Check

Anita deletes a view in a Fabric warehouse that a semantic model uses as its data source. Which downstream items might break?

AOnly the warehouse itself — views are internalBThe semantic model and all reports or dashboards connected to itCOnly reports that directly query the warehouseDNothing breaks — the semantic model caches the data

Check Answer

🎬 Video coming soon

---

Next up: XMLA Endpoint & Reusable Assets