What's the best practice for primary keys in tables?

Question

When designing tables, I\'ve developed a habit of having one column that is unique and that I make the primary key. This is achieved in three ways depending on requirements

Answer 1

Here are my own rule of thumbs I have settled on after 25+ years of development experience.

• All tables should have a single column primary key that auto increments.
• Include it in any view that is meant to be updateable
• The primary key should not have any meaning in the context of your application. This means that it should not be a SKU, or an account number or an employee id or any other information that is meaningful to your application. It is merely a unique key associated with an entity.

The primary key is used by the database for optimization purposes and should not be used by your application for anything more than identifying a particular entity or relating to a particular entity.

Always having a single value primary key makes performing UPSERTs very straightforward.

Use additional indices to support multi-column keys which have meaning in your application.