I gave a quick talk about DB schemas yesterday during the October Human Talks session as well as during BDX I/O

tl;dr:

• Design with querying in mind
• Have a look at Normal Forms
• don’t let the ORM write the schema
• use postgresql fully
• consider using UUIDs as PKs
• don’t fear joins because of MySQL poor implementation
• be consistent in PKs and FKs naming
• fields should be not null by default
• avoid deleting rows
• use enums
• use the expressive types available in PGSQL
• give jDbT a try: https://github.com/divarvel/jdbt

Download the slides in HTML.

The source files are on GitHub