Database Merger for Sequences

Question

Dear Gurus,
I have something upcoming task which include a merger of two database instances. The schema structure is exactly same and there is no duplicity problem. I want to dump the both schema data in one instance. The only challenge i have here is "How to manage the sequences?".
Explanation: Instance A.
Sequence : 1,2,3,4,....20
Instance B
Sequence: 1,2,3,4....20
Requirement : when i merge it should be
Instance C: 1,2,3,4...20 [from Instance A ] cont.. 21,22,23,24.....40 [From Instance B]
Javed A. Khan
Edited by: user8195422 on Feb 10, 2012 3:29 PM

Answer 1

The advice you got:
>
"Find the current highest value of each sequence. Set the new sequence to start with the highest of those two values plus one.
Not complicated."
>
will work for new data that is entered after the MERGE.
So the only issue you need to deal with is the exampe with INVOICE that Billy gave you. If both A and B have invoices with an ID = 123 then you can't just merge the data. For those invoices you will need to reassign the keys used by A (or B - your choice) to values that do not conflict with the other.
You need to reassign the keys in the CHILD tables also.
This type of KEY management is often done as part of the ETL process of the entire MERGE operation.
1. Create a table containing the list of ID values that are common to A.INVOICE.ID and B.INVOICE.ID
CREATE TABLE DUP_INVOICE_ID AS
SELECT ID, 0 NEW_ID FROM A.INVOICE WHERE EXISTS (SELECT 'X' FROM B.INVOICE WHERE A.INVOICE.ID = B.INVOICE.ID)
-- put the smaller table first so if B.INVOICE has fewer records use 'SELECT ID FROM B.INVOICE ...'
-- now the DUP_INVOICE_ID table has the invoices that need to get new ID values
2. update the DUP_INVOICE_ID table with the new ID values
A. get the sequence value you were going to use above (highest value of the two existing sequences and add it to the invoice ID values in the table.
for example UPDATE DUP_INVOICE_ID SET NEW_ID = ID + highSeqValue + 1;
3. Create an index on DUP_INVOICE_ID (ID) column
4. Create a new ETL staging table to hold A.INVOICE records (B if you switched tables above) - when you create this table you will query the B.INVOICE
records and if the ID value is in the DUP_INVOICE_ID table you will use the NEW_ID for the key field otherwise you will just copy the record.
5. After you have replaced all of the duplicate keys for INVOICE and its child table records you can create the indexes on those tables.
Does that help explain the process you will need to use?