When should I embed data or reference it?

Definition:

Embedding stores related data inside one document, while referencing stores a link to data in another document. Embedding makes related reads and single-document atomic updates convenient, but can duplicate data or create oversized documents. Referencing supports independently managed data at the cost of additional reads or joins.

Key points:

• Embed small related data that is normally read together, such as an order address.

• Reference large, shared or independently changing data, such as products.

• Avoid arrays that grow forever; one document has a 16 MiB size limit.

• Ask whether related records grow without a bound, change independently or must be updated together.

Example:

const order = {
  customerId: 'customer-42',      // Reference to an independent record.
  shippingAddress: { city: 'Surat', postalCode: '395001' } // Snapshot.
};

• The order keeps its delivery address even if the customer later edits their profile.

Interview point:

• Base the choice on read/write patterns and lifecycle, not a rule that MongoDB always prefers embedding.