Need one solution

Question

Hi All,
We have one view and two tables --JOB_STAGE and JOB_MAIN
View and two tables are having structure like
Job_Name varchar2(50);
JOB_QUANTITY number;
Root_id            number;
We need to add all the records into the Main table(JOB_MAIN) from the view(JOB_VIEW ) for root ids which are present into the stage(JOB_STAGE).
We have tried like : - insert into JOB_MAIN select * from JOB_VIEW where root_id in (select root_id from JOB_STAGE)
This insert statement is working fine for few number of records but we need to process one root Id at a time to avoid server performance issue in case of large number of records.
We have created one procedure to process the single root id at a time.We are calling cursor of the view for each root Id.
The procedure is executing through the job which runs at every one hour.
At the time of job execution if new data is added to the view then it is causing data mismatch.
We want to avoid the data which has been added to the view when the job is runnig. Newly added data should be inserted into the Main table in next job run.
Please provide the help.
create or replace
PROCEDURE         TMP_LOAD_JOB_MAIN
IS
v_sqlerrm        VARCHAR2(250);
errors             NUMBER;
dml_errors       EXCEPTION;
L_CNT            NUMBER := 0;
J                    varchar2(25);
cursor CUR_COMP_PATH is
         SELECT DISTINCT ROOT_ID FROM JOB_STAGE;
V_BOM_LAST_UPDATED_DT varchar2(30);       
cursor CUR_GIEA_DATA(I_STAGE_RT in number) is
        select  JOB_NAME,
                JOB_QUANTITY,
                   root_id
          from JOB_VIEW
          where ROOT_ID=I_STAGE_RT;                      
BEGIN
begin        
        OPEN CUR_COMP_PATH;
        LOOP
         FETCH CUR_COMP_PATH  INTO J;
          begin
                DELETE  FROM JOB_MAIN
                WHERE ROOT_ID = J;                                                               
            OPEN CUR_GIEA_DATA(J);
                Loop
                    Fetch Cur_Giea_Data Bulk Collect Into Bom_Order_Dt Limit 1000;
                          BEGIN
                                FORALL I in 1..BOM_ORDER_DT.COUNT SAVE exceptions
                                insert into JOB_MAIN values BOM_ORDER_DT(I);                                                               
                          end;
                       EXIT WHEN CUR_GIEA_DATA%NOTFOUND;
                END LOOP;
             CLOSE CUR_GIEA_DATA;
             COMMIT;      
            END;
            EXIT WHEN CUR_COMP_PATH%NOTFOUND;
            END LOOP;
          CLOSE CUR_COMP_PATH;
      COMMIT; 
      EXCEPTION
         WHEN OTHERS
         THEN
         ROLLBACK;
         v_sqlerrm := SQLERRM;
          rollback;
END;         
END TMP_LOAD_JOB_MAIN;

Answer 1

> This insert statement is working fine for few number of records but we need to process one root Id at a time to avoid server performance issue in case of large number of records.
What server performance issue? Is that an assumption or you are actually having performance issue?
> We have created one procedure to process the single root id at a time.We are calling cursor of the view for each root Id.
This is like putting a bullet in your head because you have stomach pain. The DML in cursor (even using bulk operation) does not scale very well. It will leave your data base in an inconsistent state (Unless you manually maintain consistency). NEVER take this approach. Its always (99.99%, 0.01 I give it as benefit of doubt) wrong.
If you have performance issue with your INSERT statement then address that. I would suggest you read Re: 3. How to  improve the performance of my query? / My query is running slow.