Please see (How Does The Batch Size Affect The Number Of Parallel Update Units [ID 1311402.1]).
parallel updates means,
worker 1 sid- 400
worker 2 sid- 401
worker 3 sid- 402
worker 4 sid- 403
while 1 worker 1 doing updates to table its having lock, how come worker 2 will do updated to same tables? it should wait for worker 1 to release lock right.
Parallel updates doesn't necessarily mean update on the same object (table) by different workers, and it means more than one worker is running different scripts (updating different data in patches). If the same script is processed by many workers then adpatch is smart enough to handle the locks on the tables.
thanks for info...
I have attended class in Oracle Internal Training rooms.
The concept they told was, If query starts running, if execution plan was created already then creating profiles or running stats will effect for next run of query, not for present run.
select status, sum(rows_processed), count(*)
where update_id = (
where script_name like '&a')
group by status;
with the above script we generally monitor adpatch work status.
while adworker running this script "ar120lclcsu.sql", we see its process batches.
We observe the batches moving slowly. with help of sqltrpt recommendations, we created profile. after creating immediatly WITHOUT restarting of job, batches moved very fastly.
My doubt is, scripts started and some batches were processed means its already started and execution plan for the script is generated, after immediatly creating profile and without restart of job how profile taken effect?
Plz help me on this.