Data Modeling for Sync Gateway

Mobile sync adds constraints that don't exist in pure server-side modeling. The Sync Function sees every document write and makes routing decisions based on document fields — so your schema must carry the metadata the Sync Function needs.

The core difference from server-side modeling

On the server, embed aggressively to avoid JOINs. For sync:

• Documents are the unit of sync — the entire document is sent to the device on every change
• Channels are the unit of access control — a document belongs to channels; a user gets channels
• Conflicts happen at the document level — large embedded arrays create large conflict surface
• Document size affects sync performance — keep documents small and focused

Rule of thumb: If two users could independently modify the same document, or if a document would grow unboundedly, split it.

Required fields for Sync Gateway

Every document that syncs needs fields the Sync Function can act on: