NoSQL databases overview

In short: NoSQL is an umbrella term for databases that do not rely on tables with a fixed schema. There are four main types: document, key-value, column and graph.

In more detail: Relational databases are strong at relationships and consistency. NoSQL systems often trade some of that for easier scaling, flexible data shapes or very high speed.

In Depth

• Document: JSON-like documents (MongoDB, CouchDB). Good for nested, changing data.
• Key-value: Fetch a value by key (Redis, DynamoDB). Very fast, ideal for caches and sessions.
• Column: Data grouped by columns (Cassandra, HBase). Strong for huge write volumes.
• Graph: Nodes and edges (Neo4j). Ideal for networks such as friendships or recommendations.
• Trade-off: No NoSQL is “better”; it depends on access patterns. For invoices and orders, SQL usually remains the safe choice.

See also: CAP theorem