If you confirm that the performance bottleneck is not caused by execution plans or resource capacity issues during the execution, it is likely caused by internal wait events. In an internal wait event, the worker threads wait for the required resources, such as:
Locks
Latches
I/O resources
Memory
You can query the V$OB_SQL_AUDIT view. If you can find SQL queries that result in a large value of the EXECUTE_TIME field, and the execution duration is prolonged mainly due to the increase of the value of TOTAL_WAIT_TIME_MICRO, the performance bottleneck is caused by internal wait events. The value of TOTAL_WAIT_TIME_MICRO is the sum of the values of several parameters, such as APPLICATION_WAIT_TIME, CONCURRENCY_WAIT_TIME, USER_IO_WAIT_TIME, and SCHEDULE_TIME. You can perform further analysis to determine the types of resources that the worker threads are waiting for.
Note
For more information about the V$OB_SQL_AUDIT view, see V$OB_SQL_AUDIT (Oracle mode) or V$OB_SQL_AUDIT (MySQL mode).