I want to partition a table that currently has an appropriate column as the primary key.

The table has a date column as well and I want the partition function to use that as its key.

I understand this unavoidably means I will have to recreate the table’s primary key with a new one that uses both the original PK column plus the date column.

It strikes me that this destroys the utility of the PK.

Theoretically a duplicated value in the original single PK column would now be permitted with the new PK.

This bothers me. Is it just the way it is? No alternative? (Apart from having to add a unique constraint/index on the original PK column?)