Instead of creating a seperated join table for one many-to-many relationship, I will design a centralized table, called 'Collection'. It will contain whole relationship data for all tables in database. This table like that:
- Towards row 1, since table A and table B has many-to-many relationship, so the value of 'Unique_key' column will be NULL
- Towards row 2 and 3, since table A and table C has one-to-many relationship, so there're multiple row value.
My question is: Compared to traditional design using a seperated join table for each of many-to-many relationships, by this way of using centralized table contains whole relationship data for all tables in database. Which way should should use?