Updating 30 million recs

Question

Hi friends,
I need some advice and help in updating records as huge as 30 million.
This is the design:
I have a staging table X with 20 million recs and a final table Y with 30 million records.
Every month I need to do a full copy from X to Y i.e. take all 20 million recs and update or insert into table Y. Everyday I do an incremental copy which is update of a few thousand recs. I use a merge statement to do this and I use version 10.2
My problem is that the full copy takes more than a day...and sometimes never completes. I tried a bulk collect and then a merge with LIMIT 1000 but this takes 23 hours to complete.
Could someone please advice what is the best approach....Note that my final table has 30 million records and all indexes in place.
Thanks for the help

Answer 1

This is just a thought.  You may have to give it a try and see if the approach below works –
1.  Rename the final table Y   to Y_Bak          -- These 2 steps would probably take a couple of secs to execute
2.  Rename the staging table X to Y
     After step 1 and 2, You need not worry about “Updates” from the stage table to the final table
       -     At this point,  rows which are present in final table Y_Bak  and not existing in the staging
              table X needs to be re-populated back into final table Y
3.     Write a small script to do this ( To handle ONLY INSERTS )–
     You may want to try disabling the constraints before running the below INSERT and enable them once the INSERT is successfully completed.
                 Declare
     COLLECTION
     CURSOR
                Begin
                        BULK COLLECT into COLLECTION LIMIT 1000;
                        FORALL i in COLLECTION.START..COLLECTION.END
                           INSERT INTO Y COLLECTION(i);
                 COMMIT;
                 EXCEPTION
                       WHEN OTHERS THEN NULL;
          End;
Shailender Mehta