You plan to use a Fabric notebook and PySpark to read sales data and save the data as a Delta table named Sales. The table must be partitioned by Year and Quarter. You load the sales data to a DataFrame named df that contains a Year column and a Quarter column. Which option describes the correct approach to write this as a partitioned Delta table?

You plan to use a Fabric notebook and PySpark to read sales data and save the data as a Delta table named Sales. The table must be partitioned by Year and Quarter. You load the sales data to a DataFrame named df that contains a Year column and a Quarter column. Which option describes the correct approach to write this as a partitioned Delta table?

• A. Save as a Delta table partitioned by Year and Quarter using save to Tables/Sales
• B. Save as a Delta table partitioned by Year and Quarter
• C. Save as a Parquet table partitioned by Year and Quarter
• D. Save as a Delta table without partitioning

Answer: (B)

When writing a Delta table, you can improve query performance by partitioning the data on the columns you commonly filter on. By partitioning on Year and Quarter, Spark will organize the data into subdirectories for each year-quarter combination, enabling partition pruning and faster scans. To achieve this with PySpark in Fabric, write the DataFrame using Delta format and specify the partitioning columns, then save as a table named Sales. For example: df.write.partitionBy("Year", "Quarter").format("delta").saveAsTable("Sales") This approach guarantees the data is stored as a Delta table and physically partitioned by Year and Quarter, meeting the requirement. Writing as Parquet would not create a Delta table, and partitioning without Delta formatting wouldn’t satisfy the need for a Delta table. Saving without partitioning would miss the required optimization.

When writing a Delta table, you can improve query performance by partitioning the data on the columns you commonly filter on. By partitioning on Year and Quarter, Spark will organize the data into subdirectories for each year-quarter combination, enabling partition pruning and faster scans.

To achieve this with PySpark in Fabric, write the DataFrame using Delta format and specify the partitioning columns, then save as a table named Sales. For example:

df.write.partitionBy("Year", "Quarter").format("delta").saveAsTable("Sales")

This approach guarantees the data is stored as a Delta table and physically partitioned by Year and Quarter, meeting the requirement. Writing as Parquet would not create a Delta table, and partitioning without Delta formatting wouldn’t satisfy the need for a Delta table. Saving without partitioning would miss the required optimization.