Data model load - Best Practice for Foreign Key Join Error Due to Duplicates?

Hi ​@el.mostafa.janah, 

In your example, I think you want to join EKKO (PO Header) to LFA1 (Vendors)? 

In this case, the (MANDT, LIFNR) combination should be unique in the Vendor table and the second scenario to remove the duplicates is preferred. 

How to remove this depends on your situation: 

• If they are real duplicates (all fields are 1 on 1 the same):
  • Select ‘Remove Duplicates during Extraction’ option in the Extraction Job (when available)
  • Use SELECT DISTINCT instead of SELECT when querying the table.
  • Instead us using various SELECT DISTINCTS in objects/events, you can create a intermediate table (e.g. LFA1_UNIQUE) and perform this operations once, and query this intermediate table in your other scripts. 
  • Note: DISTINCT is quite an expensive operation, so use only when needed: https://docs.celonis.com/en/etl-engine-best-practice.html#UUID-93423d96-20c0-f099-6165-531e23cdfe79_section-id235394097634163
• If the duplicate rows are not fully the same, you can add other logic to determine which record you want to keep. 
  • Use ROW_NUMBER() and add your logic to the ordering clause to keep the ones you want: https://docs.celonis.com/en/etl-engine-best-practice.html#UUID-93423d96-20c0-f099-6165-531e23cdfe79_section-id235394126882573