Schema Design Tips

Model for your queries

Design around how you read and write data, not only how entities relate on a whiteboard.

Embed when data travels together

• User profile settings loaded with the user
• Order line items usually shown with the order
• Small, bounded arrays

Reference when shared or large

• Authors shared by many books
• Huge comment histories better as separate collections
• Data updated independently on different schedules

Avoid unbounded growth

A document that grows forever (append-only logs inside one doc) becomes hard to update and transfer. Cap arrays or move history out.

// Prefer referencing for large shared data
{ "title": "Book", "authorId": ObjectId("...") }