I have a requirement to archive a large volume of data from a table (T1) within my published database. However, I want the archived data to remain visible and available on the subscriber databases (of which there are a very large number, each receiving a discrete partition of the published db's data).
There are no updates to T1, only inserts and deletes. The inserts are driven by selection criteria held in other tables that I will not be removing data from. The selection process populates a set of records into the table to be archived. Deletes only occur when a selection is re-run, as the current T1 set is removed and a new set is inserted.
I am thinking of disabling the replication triggers on T1, removing the data, then re-enabling the triggers. Is this a bad idea? My primary concern is what happens if a selection is re-run and delete actions are triggered for records that no longer exist at the publisher. Will this cause issues when the subscriber replicates?
Is there anything else that I should be thinking about here?
Selections can be run at both subscriber and publisher. Replication is MERGE.