Nonrelational

Nonrelational, or non-relational databases, are database management systems that do not rely on the traditional table-based relational model for data storage and operations. The term NoSQL is often used to describe these systems, though some prefer to emphasize the nonrelational nature. They emerged in the early 2000s and gained prominence with the needs of large-scale web applications requiring flexible schemas and horizontal scalability.

These databases typically store data in models such as key-value pairs, document collections, wide-column families, or graphs. They generally prioritize schema flexibility, distributed operation, and high throughput. Many are designed to scale out across commodity hardware and to tolerate network partitions, sometimes offering eventual consistency rather than immediate ACID transactions. While some systems provide transactional guarantees, the level of consistency and isolation varies.

The major categories include key-value stores (e.g., Redis, Riak), document stores (e.g., MongoDB, CouchDB), column-family stores (e.g., Cassandra, HBase), and graph databases (e.g., Neo4j, ArangoDB). Cloud-based services such as DynamoDB, Cloud Firestore, and Cosmos DB offer managed nonrelational options. Use cases include content management, real-time analytics, session stores, IoT, and applications requiring flexible data models or rapid scale.

Compared with relational databases, nonrelational systems trade some aspects of data integrity and complex querying for scalability, performance, and flexible schemas. They are often used in polyglot persistence architectures, where multiple data stores are used to fit different application needs. Criticisms include limited join capabilities, evolving data models, and varying maturity of tools and ecosystems.