Question

You are building a Fabric lakehouse table named SalesTransactions that will hold several years of order data. Analysts almost always filter their queries by the OrderDate's year and month. Using a Spark notebook, you write the Delta table with df.write.format("delta").partitionBy(...).saveAsTable("SalesTransactions"). Which partitioning strategy will best improve query performance without creating an excessive number of small partition folders?

Accepted Answer

Partition by derived Year and Month columns extracted from OrderDate.

Answer

Partition by the raw OrderDate column so every distinct date becomes its own partition.

Answer

Partition by the OrderId primary key to guarantee even data distribution.

Answer

Do not partition; instead sort the DataFrame by OrderDate before writing.