Key-Value, Document, Wide-Column, and Graph Models in Depth
Study the major NoSQL models carefully so you can choose the right one for a workload instead of treating NoSQL as a single style.
Inside this chapter
- Key-Value Databases
- Document Databases
- Wide-Column Databases
- Graph Databases
Series navigation
Study the chapters in order for the clearest path from NoSQL basics to advanced distributed design and production decision-making. Use the navigation at the bottom of each page to move through the full series.
Key-Value Databases
Key-value systems store data as a value accessed by a unique key. They are often simple, fast, and excellent for caching, sessions, feature flags, token storage, counters, and quick lookup workloads. Their simplicity is a strength, but they are usually not ideal for rich relationship-heavy querying.
Document Databases
Document databases store structured documents, often JSON-like, with nested fields and arrays. They are excellent when records naturally belong together as documents, such as product catalogs, user profiles, content items, or flexible metadata models.
Wide-Column Databases
Wide-column systems like Cassandra emphasize partitioning, distributed writes, and query-driven storage. They are commonly used for time-series data, event ingestion, and globally distributed operational systems that need very high scale.
Graph Databases
Graph databases model nodes and edges directly, which is ideal for social networks, fraud relationships, recommendation paths, dependency maps, knowledge graphs, and other workloads where relationships are the central data story.