I had a transactional database that held transactions for 6 european countries.

I split the transaction table into 12 separate tables. For each country I had

• An "Active" table (because I figured out that 90% of queries were concerned with data in the past 2 quarters).
• An "Archive" table for data older than that

My understanding with the users of the database was that they would accept slower queries on archived data.

Prior to leaving I was looking at splitting the archive tables into separate tables by year.