When a mapping contains a Filter transformation on a partitioned column of a Hive source, the Blaze engine can read only the partitions that contain data that satisfies the filter condition. To push the filter to the Hive source, configure the Filter transformation to be the next transformation in the mapping after the source.