Microsoft DP-600 Practice Exam

• A. Add a colleague as another Viewer
• B. Run SELECT queries on its tables Correct answer
• C. Rename the warehouse in the workspace
• D. Create a table by running CREATE TABLE

Option B is correct because the Viewer role grants read access to warehouse data, so the user can run SELECT queries through the SQL endpoint. Viewers cannot manage workspace access (A), rename or otherwise modify items (C), or run DDL such as CREATE TABLE (D); those need Contributor or higher.

• A. A deployment pipeline for each business area
• B. A separate capacity for each business area
• C. A Fabric domain for each business area Correct answer
• D. A managed private endpoint for each area

Option C is correct because a Fabric domain groups workspaces by business area and lets domain admins and contributors delegate ownership and governance of those workspaces. Deployment pipelines (A) promote content between stages, a capacity (B) provides compute, and a managed private endpoint (D) secures network access; none of them group workspaces by owner.

• A. Query insights in the Fabric warehouse
• B. Performance Analyzer in Power BI Desktop
• C. Impact analysis in the lineage view Correct answer
• D. The Monitoring hub

Option C is correct because impact analysis in the lineage view lists the reports and other items, including those in other workspaces, that depend on a semantic model. Query insights (A) reports warehouse query activity, Performance Analyzer (B) times report visuals in Desktop, and the Monitoring hub (D) tracks item activity runs, not dependencies.

• A. The XMLA endpoint of the workspace Correct answer
• B. The Power BI REST API datasets endpoint
• C. The SQL analytics endpoint of the lakehouse
• D. The OneLake ABFS path of the model

Option A is correct because Tabular Editor and other external tools connect to a published semantic model through the workspace's XMLA endpoint. The REST API (B) is not a Tabular object model connection, and the lakehouse SQL analytics endpoint (C) and OneLake path (D) expose lakehouse data, not the model's metadata.

• A. Share the lakehouse with Apache Spark read access
• B. Share the lakehouse with SQL endpoint read access only Correct answer
• C. Grant Build on the lakehouse's default semantic model
• D. Add the analyst as a workspace Viewer

Option B is correct because sharing the lakehouse with only the SQL endpoint read permission lets the analyst query its tables with T-SQL without access to the files or other items. Spark read access (A) targets files and notebooks, Build on the default semantic model (C) does not give T-SQL access, and a workspace Viewer role (D) grants access to every item in the workspace.

• A. A scalar function called from a CHECK constraint
• B. A database role that has been granted SELECT on the table
• C. A DAX table filter role on the default semantic model
• D. A security policy with a filter predicate on the table Correct answer
• E. An inline table-valued function that returns a row for each permitted row Correct answer

Options D and E are correct because row-level security in a Fabric warehouse is built from an inline table-valued function that returns a row only when the predicate allows it, plus a security policy that applies that function as a filter predicate on the table. A CHECK constraint does not filter reads, a database role with SELECT only grants access to the whole table, and a DAX role on the default semantic model does not affect T-SQL queries.

• A. The Delta files behind the table store the values in masked form
• B. The stored values in the table stay unchanged Correct answer
• C. Only members of the db_datareader role see masked values
• D. Users without the UNMASK permission see the masked form in query results Correct answer
• E. A WHERE clause on the column can no longer reveal the real values

Options B and D are correct because dynamic data masking changes only what query results show: the stored values are untouched, and users who lack UNMASK (and are not Admin, Member or Contributor in the workspace) see the masked form such as aXXX@XXXX.com. Option A is wrong because the stored data is not masked. Option C is wrong because masking is not tied to db_datareader membership. Option E is wrong because a WHERE clause on a masked column can still be used to infer the real values.

• A. Mirrored database
• B. Lakehouse Correct answer
• C. Fabric Data Warehouse
• D. Eventhouse

Option B is correct because a lakehouse stores unstructured and semi-structured files in its Files area and Delta tables in its Tables area within a single item, and Spark notebooks work on both. A mirrored database and an Eventhouse are not designed to hold arbitrary image files, and a warehouse is a T-SQL engine over tables rather than a file store.

• A. Enable V-Order on the shortcut so it registers as a table
• B. Create it under Files and load the CSV to a Delta table Correct answer
• C. Rerun the metadata sync on the SQL analytics endpoint
• D. Grant Contributor on the ADLS Gen2 storage account

Option B is correct because shortcuts in the Tables section are only recognized as tables when the target holds Delta data, so a folder of raw CSV files is never registered. Creating the shortcut under Files and loading the CSV into a Delta table produces a real table. V-Order (A), a metadata sync (C) and storage permissions (D) do not convert CSV files into Delta tables.

• A. Create a OneLake shortcut in the lakehouse to the mirrored tables Correct answer
• B. Connect the notebook to Azure SQL Database over JDBC
• C. Schedule a Dataflow Gen2 that loads the mirrored tables into the lakehouse
• D. Run INSERT INTO through the mirrored database's SQL analytics endpoint

Option A is correct because a OneLake shortcut in the lakehouse points at the mirrored Delta tables in place, so the notebook reads them next to lakehouse tables with no second copy. JDBC to the source and a scheduled Dataflow Gen2 both re-read or replicate the data, and a mirrored database's SQL analytics endpoint is read-only, so INSERT INTO is not an option.

• A. Deployment pipelines
• B. Workspace lineage view
• C. Real-Time hub
• D. OneLake catalog Correct answer

Option D is correct because the OneLake catalog lets users browse and discover items across workspaces and filter them by item type, owner and endorsement. Deployment pipelines promote content between stages, the lineage view shows dependencies inside one workspace, and the Real-Time hub is for discovering streaming data.

• A. Spark notebook reading over JDBC
• B. Data pipeline with a Copy activity Correct answer
• C. Eventstream with a SQL Server source
• D. Dataflow Gen2 loading to the warehouse

Option B is correct because a data pipeline Copy activity is built for high-volume, no-transformation bulk copies and can reach on-premises SQL Server through an on-premises data gateway. A Spark notebook over JDBC and Dataflow Gen2 add overhead for a plain copy, and an Eventstream is for streaming events, not nightly batch loads.

• A. Merge the queries with a left outer join on Date
• B. Table.Combine({Sales2024, Sales2025}) Correct answer
• C. Append the two queries into a new query Correct answer
• D. Pivot the Date column with a Sum aggregation
• E. Table.NestedJoin(Sales2024, {"Date"}, Sales2025, {"Date"}, "Rows")

Options B and C are correct because appending (the Append queries step in the Dataflow Gen2 editor, or Table.Combine in M) stacks the rows of two tables that have identical columns into one table. Merging with a left outer join or Table.NestedJoin matches rows side by side and adds columns instead of rows, and pivoting the Date column reshapes the data without combining the two sources.

• A. SELECT CustomerID, Name, City FROM dbo.StgCustomer GROUP BY CustomerID, Name, City Correct answer
• B. SELECT * FROM dbo.StgCustomer WHERE CustomerID IS NOT NULL
• C. SELECT TOP 1 * FROM dbo.StgCustomer ORDER BY CustomerID
• D. SELECT DISTINCT CustomerID, Name, City FROM dbo.StgCustomer Correct answer
• E. SELECT CustomerID, COUNT(*) FROM dbo.StgCustomer GROUP BY CustomerID

Options A and D are correct because SELECT DISTINCT over all three columns, and a GROUP BY on all three columns with no aggregates, both collapse exact duplicate rows into one. Filtering out null CustomerID values removes no duplicates, TOP 1 returns a single row, and grouping by CustomerID alone only counts rows per customer.

• A. A one-to-many relationship links the parameter table to the fact table
• B. A calculated table is added to the model Correct answer
• C. A calculation group with one calculation item per field is added
• D. Each field is a row with a display name, a NAMEOF reference and a sort order Correct answer
• E. A hidden measure that uses SWITCH is generated for each field in the parameter

Options B and D are correct because a field parameter is created as a calculated table whose rows each hold a display name, a NAMEOF reference to the field and a sort order. Desktop does not create a relationship to the fact table (A), a calculation group (C), or a generated SWITCH measure (E); the visual simply binds to the parameter's fields.

• A. CALCULATE([Sales], TOPN(1, ALLSELECTED(Product[Name]), [Sales], DESC))
• B. CALCULATE([Sales], OFFSET(1, ALLSELECTED(Product[Name]), ORDERBY([Sales], DESC)))
• C. CALCULATE([Sales], OFFSET(-1, ALLSELECTED(Product[Name]), ORDERBY([Sales], DESC))) Correct answer
• D. CALCULATE([Sales], OFFSET(-1, ALLSELECTED(Product[Name]), ORDERBY(Product[Name])))

Option C is correct because OFFSET(-1, ..., ORDERBY([Sales], DESC)) returns the row one position before the current one in the sales-descending order, which is the product listed directly above. ALLSELECTED keeps the order within the Region slicer's filter. Option B looks one row below, option D orders by name so the 'row above' is alphabetical, and option A's TOPN always returns the single top product.

• A. CALCULATE([Sales], OFFSET(-2, ALLSELECTED(Date[MonthStart]), ORDERBY(Date[MonthStart])))
• B. AVERAGEX over WINDOW(-2, REL, 0, REL) on product-month rows that have sales, partitioned by product Correct answer
• C. CALCULATE([Sales], DATESINPERIOD(Date[Date], MAX(Date[Date]), -3, MONTH)) divided by 3
• D. AVERAGEX over WINDOW(-2, REL, 0, REL) on every month in the Date table, with no partition by product

Option B is correct because WINDOW(-2, REL, 0, REL) over only the product-month rows that have sales, partitioned by product, spans the current and two previous months with sales for that product, and AVERAGEX averages them. Options A and C use calendar months, so gaps without sales break the requirement, and option D includes every month and ignores the product partition.

• A. Compute the column in the Delta table and refresh the model Correct answer
• B. Add a Power Query step to the table in the semantic model
• C. Add a DAX calculated column to the table in model view
• D. Change the table to Dual storage mode and add a calculated column

Option A is correct because Direct Lake tables on the SQL analytics endpoint do not support DAX calculated columns or Power Query steps, so the column has to be computed upstream in the Delta table and picked up on refresh. Options B and C are not supported on a Direct Lake on SQL analytics endpoint table. Option D does not fit because the table must stay in Direct Lake mode, and Dual is a different storage mode that cannot be mixed with Direct Lake tables in the same Direct Lake on SQL model.

• A. MAX(Sales[Amount])
• B. MAXX(VALUES(Sales[OrderDate]), [Sales])
• C. MAXX(ALL('Date'[YearMonth]), [Sales])
• D. MAXX(VALUES('Date'[YearMonth]), [Sales]) Correct answer

Option D is correct because VALUES('Date'[YearMonth]) returns only the months visible under the Year slicer, and MAXX evaluates [Sales] for each month and returns the highest. Option C uses ALL, which ignores the slicer, option B iterates order dates rather than months, and option A returns the largest single transaction amount.

• A. Use an explicit measure for Quantity in those visuals Correct answer
• B. Mark the Date table again as the model's date table
• C. Raise the precedence value of Time Calc above the other groups
• D. Set the Summarize by property of Quantity to Sum

Option A is correct because calculation items apply only to explicit measures; a column dragged into a visual and summed implicitly is not changed by the calculation group. Creating an explicit measure for Quantity lets the YTD item apply. Marking the date table, raising precedence, or changing Summarize by does not make an implicit measure respond to a calculation item.