Its just an ETL table and I can afford to redo it in case of error
coz i got a backup copy.
I am converting a Foxpro database to Oracle. Its a Loans Payment table.
I got 180million plus Loans Payment foxpro table,
the tables reside on different branches(decentralized),
and its branches have different structure of that table ( what a
delimma) because each branch has its own programmer,
with different style to maintain the apps.
I have to merged it in the central office.
I built a common structure table that fits for all the branches.
Then I extract using HS connectivity.
I start cleaning it up ...updates...updates...updates...
yeah the parallel mention by tamil may help.
I also partitioned it with the loan_date per year.
One of my delimma is there are lots of erroneous loan_date. So
I cannot load it into the partitioned table now. But the management
tell me that all the records must be accounted for ...that all of it must
be inquired (seen in a inquiry form) even if errorneous as it is.
How can I merged this erroneous date rows to the main partitioned table?
I am loading the erroneous loan_date rows to another table then make
a VIEW to join the two. Is this the right way hunnies?
I think you are building the database for the first time and the big update is going to be one time job. And you wanted to update the table in a shorter time.
1 Take DB backup.
2 Change the underscore parameter, _disable_logging to TRUE.
3 Drop indexes on the table and Update the table using parallel options.
4 Verify the updated columns in the tables.
5 Shutdown the DB and change the parameter, _disable_logging to FALSE.
6 Startup the DB. Verify integrity of the DB (all the tables and indexes).
7 Take a backup now.
In case some thing goes wrong during the update, restore the backup copy taken in step-1.