The current setup:
A live application DB is on an oracle server, which has a nightly replication over to an SQL server that users (read me) can use to bash out reporting from.
The issue is due to the server's existing on different sites the job is now taking far too long to complete and there is a need to reduce the volume of data being copied to the SQL server daily.
The goal:
To enable archiving to a new SB in SQL server the aim is to replicate a rolling 7 days (ideally) rather than the entire DB and the archiving can all be managed SQL Server side. This allows correction to be made on the Ms side should the archiving ever fail when there is no support available for a day or two.
The issue is most of the tables have no date/time fields, but all of them do have unique ID fields.
I am looking for all manner of ideas the forum can conjur up that would manage this.
Safe things to be assumed: cannot have the live server update the SQL server as part of the transaction, this needs to be done as some form of replication job.
It might but also might not be possible to check values in the SQL server.
One idea I have is to make a log of the last ID number in every table at the start of each replication and store against the date. This would be stored in a table in the SQL server (cannot alter the system DB in any form to put it there.) After 7 days it would become possible to use this table to look up the ID number that you want results >=
This would work for all tables with no date/times as there are no record updates.
There are some record update tables, but these have modify date/time fields and therefore not an issue.
Many thanks in advance for thoughts on this.