Structured vs unstructured, NoSQL taxonomy

Structured vs unstructured data

Structured data follows a fixed schema of rows and columns (e.g. a bank's customer table). Unstructured data has no predefined format or schema (e.g. text, images, video, emails, logs).

(Semi-structured data such as XML and JSON has tags but no rigid schema.)

Taxonomy of NoSQL

NoSQL ("Not Only SQL") databases are non-relational, schema-flexible, horizontally scalable stores. They are classified into four main types by data model.

1. Key-Value stores: data is a pair of unique key and an opaque value, accessed by key only. Very fast and simple; good for caching, sessions, shopping carts. Examples: Redis, Riak, DynamoDB, Memcached.
2. Document stores: data is stored as self-describing documents (JSON/BSON/XML) that can be nested and differ in structure. Supports querying on fields. Good for content management and catalogues. Examples: MongoDB, CouchDB.
3. Column-family (wide-column) stores: data is stored in rows with column families, and column values are stored together. Good for huge write volumes and analytics on large data. Examples: HBase, Cassandra, BigTable.
4. Graph databases: data is stored as nodes, edges and properties, optimised for traversing relationships. Good for social networks, recommendations and fraud detection. Examples: Neo4j, InfiniteGraph.

            NoSQL
   +--------+--------+---------+
 Key-Value Document Column   Graph
 Redis    MongoDB  HBase    Neo4j
 Riak     CouchDB  Cassandra

The first three (key-value, document, column-family) are called aggregate-oriented databases, as they store a whole aggregate unit together, which suits clusters. Graph databases are relationship-oriented.