I find this summary in Eric Redmond’s ‘A Little Riak’ book very concise and useful.
Relational Traditional databases usually use SQL to model and query data. They are useful for data which can be stored in a highly structured schema, yet require lexible querying. Scaling a relational database (RDBMS) traditionally occurs by more powerful hardware (vertical growth).
Examples: PostgreSQL, MySQL, Oracle
Graph These exist for dataighly interconnected data. They excel in modeling complex relationships between nodes, and many implementations can handle multiple billions of nodes and relationships (or edges and vertices). I tend to include triplestores and object DBs as specialized variants.
Examples: Neo4j, Graphbase, IniniteGraph
Document Document datastores model hierarchical values called documents, represented in formats such as JSON or XML, and do not enforce a document schema. They generally support distributing across multiple servers (horizontal growth).
Examples: CouchDB, MongoDB, Couchbase
Columnar Popularized by Google’s BigTable, this form of database exists to scale across mul- tiple servers, and groups similar data into column families. Column values can be individually versioned and managed, though families are deined in advance, not unlike RDBMS schemas.
Examples: HBase, Cassandra, BigTable 4
Key/Value Key/Value, or KV stores, are conceptually like hashtables, where values are stored and accessed by an immutable key. They range from single-server varieties like Memcached used for high-speed caching, to multi-datacenter distributed systems like Riak Enterprise.
Examples: Riak, Redis, Voldemort