Question

A data engineer manages a large Delta table in a Fabric lakehouse that stores three years of sales transactions. Analysts run Power BI Direct Lake reports that almost always filter by a single calendar month. Queries have become slow, and file inspection shows thousands of small Parquet files across the table with no partitioning. The engineer wants to improve read performance for the month-filtered queries with minimal ongoing maintenance. What should the engineer do?

Accepted Answer

Partition the Delta table by a low-cardinality month column and periodically run OPTIMIZE with V-Order to compact files.

Answer

Partition the Delta table by the transaction ID column so every query can seek directly to a single file.

Answer

Convert the table to a non-Delta CSV format to reduce Parquet metadata overhead during reads.

Answer

Disable V-Order on the table so that writes complete faster and reduce the number of small files.