Well, the thing that I must suffer in my databases is when the developers learn how to use hint and put "parallel (32)" in every query. They probably think that this hint is a "set magic=on" that will make everything faster at no cost.
So, when you find the "over-parallelism" disease in one of your databases, is where knowing how to tune it makes really diference.
For this article, I really recommending reading and understanding the documentation referenced below. As I said in this topic, it's very important to understand every parameter that change the parallelism behaviour. Run the query below and check the contents and use for every table:
select object_name from dba_objects where object_name like 'V$P_\_%' escape '\';
I would recommend changing your session parallel parameters and running some query than checking the parallelism in V$PQ_SESSTAT.
Path to Documentation:
VLDB and Partitioning Guide -> 8 Using Parallel Execution -> Tuning General Parameters for Parallel Execution
VLDB and Partitioning Guide -> 8 Using Parallel Execution -> Monitoring Parallel Execution Performance
VLDB and Partitioning Guide -> 8 Using Parallel Execution -> Miscellaneous Parallel Execution Tuning Tips
Click here to go back to the Main OCM 11g Preparation page.
Have you enjoyed? Please leave a comment or give a 👍!