I need help to clear my understanding on few points as mentioned below.
One way of loading data into NOPI table is using Fast Load. There TD uses configuration map and list downs all AMPs including the down AMPs for loading in a round robin fashion. In short this is the bottomline of this method. For Down AMPs the records are loaded into corresponding Fallback owner. Now my question is, what will happen if the NOPI table is not defined as Fallback ? Will the loading fail or the rows will be lost ?
What is the best option to truncate load data into staging tables (tables are empty during loading) based on 1-to-1 mapping from source flat files ? I was thinking about two options,
a) NOPI table without any partition and use TPT STREAM with SERIALIZE OFF , ARRAY SUPPORT ON and PACK data as many as possible in single stmt
b) Standard PI table with Fast Load
Here the stage tables have no other index or constraints.
Experts, Please let me know your response.