How teradata makes sure that there are no duplicate rows being inserted when its a SET table? Teradata redirects the new inserted row as per its Primary Index to the target AMP (Access Module Processor) on the basis of its row hash value, and if it find same row hash value in that AMP then it start comparing the whole row, and find out if duplicate. When the target table has UPI(Unique Primary Index) then the duplicate row is rejected with an error. In case of a NUPI(Non-Unique Primary Index) then it is rejected without throwing any error