April 26, 2014 \ Ananth TM Teradata API TD API has the option to run in parallel mode also. One can write/insert records using TD API in parallel mode and it gives much better performance than using the same in sequential mode. When used in parallel mode, you must set the partitioning property to HASH and select all the columns in the Primary Index of the target table for partitioning. You might get occasional blocks due to Teradata hash synonyms only, but this should be pretty minimal on most low volume loads. If you plan to implement restart strategy in a job using TD API for loading data, since there is no cleanup, it is advised to write as UPSERT. The API will generate upsert code automatically, but it will be 2 SQL statements, not atomic upsert. If you want ATOMIC upsert you will have to take the auto generated code and then modify it slightly to the ATOMIC upsert syntax.