Microsoft DP-600: Prepare data Practice Questions

Using PySpark to cast the column to 'date' type and overwriting the Delta table updates the actual schema and stored data, making the column natively usable with date functions via the SQL analytics endpoint. Creating a view that casts the column is a valid workaround but does not fix the underlying table schema; analysts querying the base table directly will still encounter issues. ALTER TABLE for adding computed columns is not supported in the Fabric Warehouse/SQL analytics endpoint in the same way as SQL Server, and renaming the column would break existing queries. The Visual Query Editor does not modify the underlying Delta table schema; it only generates transformation steps for ad-hoc output.

Views and table-valued functions are both supported at the Lakehouse SQL analytics endpoint, allowing analysts to encapsulate reusable transformation and filtering logic that other consumers can query. Stored procedures are supported in Fabric Warehouse but are NOT supported at the Lakehouse SQL analytics endpoint, making this a common misconception. Clustered columnstore indexes are a Warehouse concept; Delta tables in a Lakehouse manage their own Parquet-based file layout and do not support explicit index creation. External tables over Azure Blob Storage are a Warehouse feature (via CREATE EXTERNAL TABLE with OPENROWSET), not natively created inside a Lakehouse through the SQL endpoint.

The 'Replace values' transformation in the Visual Query Editor allows users to specify a value to find (including null) and a replacement value, achieving the desired result without writing code. 'Remove empty rows' deletes rows with nulls rather than replacing the null with a default value. 'Filter rows where ProductCategory is not null' removes those rows entirely instead of filling them. 'Change column type' changes the data type of a column but does not replace null values.

Fabric Warehouse is the correct choice because it is a fully managed relational data warehouse that supports T-SQL DDL/DML, enforces referential integrity constraints (including foreign keys), and is designed for concurrent transactional writes from multiple applications. A Lakehouse with Delta tables supports T-SQL reads via the SQL analytics endpoint but does not enforce referential integrity or support T-SQL-based concurrent writes natively. An Eventhouse KQL database is optimized for time-series and log analytics using KQL, not relational T-SQL workloads. The Lakehouse SQL analytics endpoint is a read-only T-SQL interface over Delta tables and does not support write operations.

ALL(DimProduct) removes all filters from the DimProduct table, so the denominator computes total sales across all products regardless of any product filter in the visual — this is the classic market share pattern. ALLEXCEPT(FactSales, DimProduct) removes all filters on FactSales EXCEPT those coming from DimProduct, which is the opposite of the desired behavior and would keep the product filter. REMOVEFILTERS() with no arguments removes all filters from all tables, which may inadvertently remove date or other slicer filters, producing an incorrect denominator. KEEPFILTERS(ALL(DimProduct)) is a contradiction — KEEPFILTERS adds constraints while ALL removes them, and the combination does not achieve the desired result.