Deleting Data from 85 tables

Question

Hi All,
There is a 13 years old data in DB, which holds data from 2001. The task assigned to me is to delete the old data from Main table and its associated tables.
87 tables are identified and data needs to be removed.
1) We don't have an option to bring down the DB, the DB should be online.
2) We can't disable triggers and constratins.
3) We can't disable LOGGING
4) The total Size of the tables that we are going to do the deleting is around 290 GB.
5) DB is Oracle 11g.
What I did is,
I wrote a Stored Procedure, that get's MONTH and YEAR as input parameter. This procedure is called 12 times to delete one year.
I have collected all the ID’s and stored in a separate table. Based on the id, I am deleting the data from each table and committing. (I was doing a bulk commit after deleting some group of tables, but those commits took lot of time.).
Now total time to delete 150 Million Rows from the entire table is 4 days to delete 7 years of data.
Is it any way I can make the process faster other than adding INDEX to avoid full table scans.
is there is a better approach to handle this?.
Advance Thanks for your response.
-Jac

Answer 1

988197 wrote:
Hi All,
There is a 13 years old data in DB, which holds data from 2001. The task assigned to me is to delete the old data from Main table and its associated tables.
87 tables are identified and data needs to be removed.
1) We don't have an option to bring down the DB, the DB should be online.
2) We can't disable triggers and constratins.
3) We can't disable LOGGING
4) The total Size of the tables that we are going to do the deleting is around 290 GB.
5) DB is Oracle 11g.
What I did is,
I wrote a Stored Procedure, that get's MONTH and YEAR as input parameter. This procedure is called 12 times to delete one year.
I have collected all the ID’s and stored in a separate table. Based on the id, I am deleting the data from each table and committing. (I was doing a bulk commit after deleting some group of tables, but those commits took lot of time.).
Now total time to delete 150 Million Rows from the entire table is 4 days to delete 7 years of data.
Is it any way I can make the process faster other than adding INDEX to avoid full table scans.
If you added another index, that would likely make the operation SLOWER because every time a row is deleted, the index would also have to be changed to reflect that fact. Indexes introduce more work to be done on DML operations.
RP0428 has it right. This is a one-time operation and you are already well down the road. Just keep going.
is there is a better approach to handle this?.
Advance Thanks for your response.
-Jac