Link to home
Start Free TrialLog in
Avatar of prashubk
prashubkFlag for Singapore

asked on

MERGE and BULK Update Using Cursor

Hi Experts,

              For a huge set of data merging should i be using MERGE or Bulk update using Cursor in oracle?.How do we  decide which one is appropriate based on the situation?.

Thanks
prashu
Avatar of Ora_Techie
Ora_Techie

Since MERGE will process the sets of records, i would prefer it over bulk update. Don't go for PL/SQL if you can do it in straight SQL (think in sets).
If using of PL/SQL is necessary then use collections,fetch bulk collect and forall.
update of an inline view would be the most effective in this case.

update (select ...,...,... from tableA,tableB) set ........
Avatar of prashubk

ASKER

Thanks for the input..i have a procedure that is where i would be  taking the values from two globle temp table.Each table consists of about 300,000 records.
I have used MERGE for populating the data. I was just thinking since i'm dealing with huge data and commit will happen only at the end will it pose any issues.
If i use cursor then can fetch 10000 records at a time from the cursor  then commit after updating .Again fetch another 10000 and commit after updating as shown in the piece of code.Please suggest which is better for this kind of situation
nArraySize NUMBER := 10000;
OPEN list;
LOOP        
 FETCH list BULK COLLECT 
            INTO    xId, tCode,pCode,colOne,colTwo,colThree LIMIT nArraySize;
          BEGIN
            FORALL i IN 1 .. xId.COUNT
              UPDATE TRANS_TABLE
                SET col1 = colOne(i),
                    col2 = colTwo(i),
                    col3 = colThree(i)                    
               WHERE 
	          x_code = xId(i)
                 AND t_Id = tCode(i)
                 AND p_code = pCode(i);
            COMMIT;
	    EXCEPTION
            WHEN OTHERS THEN
              pRetCode := SUBSTR(SQLERRM, 1, 150);
          END;
          EXIT WHEN list%NOTFOUND;
        END LOOP;

Open in new window

if you are only updating 300,000 rows, then no need to split into pieces.    That's a small transaction.  Do it in SQL, no pl/sql needed or desired.
Hi Sdstuber

              Thanks, Then in when i should consider using splitting of data into small pieces. I mean transactions in numbers?.Please suggest
that depends on the size of your database + size of UNDO + size of your server
ASKER CERTIFIED SOLUTION
Avatar of Sean Stuber
Sean Stuber

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
SOLUTION
Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial