When you configure a group by field, the

mapping

task groups rows with the same data into a partition. Then, the Java code runs for each partition in the transformation. For example, the input row behavior is processed for each partition and each row in the partition, and the end of data behavior is processed for each partition after processing all rows in the partition.

When you select more than one group by field, the task creates a partition for each unique combination of data in the group by fields.

If you do not configure a group by field, the Java code runs based on the data's default partitioning scheme.

If you use a parameter for the group by fields, define the group by fields when you run the mapping or when you configure the

mapping

task.