You save a DataFrame to Parquet with partitionBy year, month, and day. Which statement about the resulting files is true?

• A. The results will form a hierarchy of folders for each partition key
• B. The results can only be read on a single node
• C. The partitions are not compressed
• D. Files cannot be read in parallel across nodes

Answer: (A)

Partitioning by year, month, and day creates a folder hierarchy where each combination of partition values becomes its own directory path, typically something like year=YYYY/month=MM/day=DD, with Parquet files inside those folders. This layout is exactly what the statement describes and is why partitioned writes organize data into a structured directory tree. Reading from such a layout can be done in parallel across nodes because the data is distributed across many files and folders, and Spark can prune irrelevant partitions to read only the needed folders. Also, Parquet files can be compressed, and partitioning itself doesn’t dictate compression.