How to map relations in large data set "real-life" scenario

For performance related tests, I'm trying to create a large amount of data (250k,500k,1mill records) and insert it into a DB, while emulating a "real-life" scenario.

I was wondering if there were studies/anyone knows how to best emulate a real-life scenario of data-relations. For example, if I had X amount of different classes, how many of them would be referencing other classes.

The only thing I'开发者_运维技巧m not sure about is how to emulate the relations (inheritence,reference, etc).

Thanks in advanced, Ben.

one-to-many relationship is usually implemented by a column on the "many" side, which identifies the "one" side.

many-to-many relationship is usually implemented by a separate table which identifies pairs of parent-child.

Based on your question, you're probably dealing with the latter case.