Separating duplicate and non-duplicate rows to separate tables

Step  1: Drag  the source to mapping and connect it to an aggregator transformation.

Step  2: In aggregator transformation, group by the key column and add a new port  call it count_rec to count  the key column.

Step  3: connect  a router to the  aggregator from the previous step.In router make two groups one named "original" and another as "duplicate"
In original write count_rec=1 and in duplicate write count_rec>1.

The picture below depicting group name and the filter conditions

Step 4: Connect two group to corresponding target table.