The easiest way would be to use DBMS_SCHEDULER to schedule a job for each row that you want to process. You must pass the key to a permanent table to identify the row you want to process, or to put the full row in the arguments for the job, since the temporary contents of the table are not displayed in different sessions. The number of parallel jobs is controlled by a resource manager, mainly a limited number of processors.
Why do you want to process line by line? Setting operations is performed in most cases much faster.
source
share