I have a Informatica mapping which does both insert and update on the same table. Currently it tables almost 4 hours to do update for almost seventeen million records. Can any one help me on determining how to perform fine tune the mapping either by using partitioning or by some other way possible. The source and target are from the same Oracle database. The join type is detail outer join. The target table does not have any DB level partitioning. It just has 4 index including the primary key column on which we are assigning as the key column to do the updates. Can anyone please suggest some tips to improved the performance?

The mapping design and session stats

Source table 1 : 9948154 rows

Source table 2 : 27243705 rows