Does anyone know why the mload delete task does not allow delete based on a primary index?
I need the answer just out of curiosity.
MLOAD DELETE is for deleting a large number of rows during a Full Table Scan while avoiding the Transient Journal.
When you need to delete based on the PI it's a quite efficient single AMP, (probably) single datablock access.
Thank you for the response.
I have always had this assumption that all restrictions with MULTILOAD are associated with the fact that multiload does not allow amps to talk to each other(which happens in case of USI's, JI's,etc) thereby saving time and resources.
If we consider the restriction to not allow delete based on primary index, there really isn't any communication among the amps.
So why does multiload have it?
Is it because of the fact that multiload applies a table level lock as part of its mechanism and it really makes no sense to apply a table level lock and do a PI based delete(hence, teradata just chose to have this restriction while developing multiload)?