Getting dupplicates records if we are using Historical Assignment UE and Performance issue.

Summary:

Getting dupplicates records if we are using Historical Assignment UE and it is taking too much time to complete Extract.

Content (required):

We are using PER_EXT_SEC_HISTORY_ASSIGNMENTS_UE as root and CMP_ASSIGNMENT_SALARY_INFO_RGE_UE as child. It is changes only extract. We defined Assignment ID as Threading group and Assignment_id, Assignment Start date and effective sequence are Key attributes. If we are running the extract it is taking more than 3hr'sto complete the extract and also we are getting the duplicate the records.

Any inputs to avoid duplicate records as well as improve performance.