Couchbase Server Data Modeling

Core Principle

Data that is accessed together should be stored together.

Design documents around access patterns, not normalized entities. Embedding eliminates JOINs and enables atomic KV operations.

Data Hierarchy

Cluster
└── Bucket (≤ 30 per cluster, ≤ 20 MB per document)
    └── Scope (≤ 1000 per cluster)
        └── Collection (≤ 1000 per cluster)
            └── Document (JSON, unique key within collection)

Keyspace path: `bucket`.`scope`.`collection`