NoSQL

NoSQL

A generation of database management systems not based on the traditional relational model.

What "NoSQL" actually stands for

"Not Only SQL"

Five NoSQL characteristics

(1) Not based on the relational model, (2) Support distributed architectures, (3) Provide fault tolerance and high scalability/availability, (4) Support large amounts of sparse data, (5) Geared toward performance over consistency.

Four main categories of NoSQL databases

Key-Value, Document, Graph, Column-Oriented.

Key-Value (KV) database

A NoSQL model that stores data as key-value pairs in which the value is unintelligible to the DBMS.

Bucket (KV database)

A logical grouping of keys, similar to a table; a key can appear only once within a bucket.

Three KV operations

Get (retrieve value by key), Store (write value to key, replacing any existing value), Delete (remove the key-value pair).

Examples of Key-Value databases

Dynamo, Riak, Redis, Voldemort.

Document database

A NoSQL model that stores key-value pairs in which the value is a tag-encoded document (XML, JSON, BSON), and the DBMS understands the document's content.

Collection (document database)

The grouping container for key-value pairs, analogous to a bucket in KV databases.

Key difference between KV and Document databases

Document DBMSs understand and can query the value's internal structure; KV DBMSs do not.

Examples of Document databases

MongoDB, CouchDB, OrientDB, RavenDB.

Graph database

A NoSQL database that uses graph theory to store entity instances and the relationships between them, represented as nodes and edges.

Node (graph DB)

A single entity instance.

Edge (graph DB)

A relationship between nodes.

Property (graph DB)

An attribute describing a node or an edge.

How is graph data physically stored

Often in structures like an adjacency matrix or as key-value pairs, even though it is visualized as nodes and edges.

Examples of Graph databases

Neo4j, ArangoDB, GraphBase, Aerospike.