Common table references, Int or String identifier?

Question

I was wondering how to take the best way to solve this situation. It's a database structure problem... Let's see the case.

I have some databases which have the same structure (let's say db_A, db_B, db_C). They are identical in structure but with different data, of course.

There's another database with permissions (db_permissions), and another with users (db_users).

The db_permissions structure table would be ID, permission; where permission is a unique string identifier (like post:create).

Now, there's a user which has different permissions on each table. For example: user 123 would have rights to write a post in db_A, rights to only read posts on db_B, and rights to read, write and delete posts on db_C.

My question is, how should I reference the permissions on each table?

• Do I reference the unique string identifier (like post:create, post:read) and add this directly to each database permissions (like user_id - permission_string)?

• Or do I reference them by the ID primary key on the permissions database (like user_id - permission_id)?

In my PHP script, when I check the permissions, I check by the unique string identifier.

What would you recommend? Would you use another structure for the permissions?

Thanks for your time and answers!

Answer 1

If you have a permissions table that maps IDs to permission strings, then you should use those IDs in other tables that store user permissions.

You should have a table_permissions table, with 3 columns: user_ID, table_ID, permission_ID.

Answer 2

Indexes / references should always be type integer if you want to keep performance