An Introduction to NoSQL Databases

Based on what I found at the official NoSQL site, here is a brief introduction combined with material from around the web.

Key-value databases

• The more widely used ones

  • redis (open source)
  • memcached (open source)

• Characteristics:

    Processes data in memory, so it is relatively the fastest; the data storage structure is the simplest, with only the key-value form; support for querying and aggregating values is very weak; because data is processed in memory, data durability is relatively weak. But redis does have big-data management capabilities; transaction support is weak.
  

  Document databases

• The more widely used ones

  • mongodb (open source)
  • couchbase (open source)

• Characteristics:

    mongodb processes data on disk and is more than ten times faster than SQL databases; couchbase processes data in memory; both have very strong horizontal scaling capabilities; the values in document databases can handle data with complex document structures, and query and aggregation performance is relatively stronger than that of key-value databases. They have big-data processing capabilities; there is no transaction support.
  

Column-family databases

• The more widely used ones

  • cassandra (open source)
  • hbase (open source)

• Characteristics:

    Processes data on disk, mainly used for big-data processing; write speed is noticeably faster than read speed, while overall read/write speed is slower than that of key-value and document databases; it has powerful data query and aggregation features; there is no transaction support.
  

Graph databases

• The more widely used ones

  • neo4j (open source in early versions)
  • orientdb (open source)

• Characteristics:

        Disk-based data processing, focused on graph (here 'graph' means graph theory in mathematics, that is, a combined system containing a number of nodes with edge relationships between them, where nodes and edges can carry associated attributes) data querying and computation. ACID transactions
  

Multi-model databases

• The more widely used ones
  • ArangoDB (open source + commercial)
  • OpenLink Virtuoso (open source + commercial)

Object databases

• The more widely used ones
  • db4o (open source)
  • versant (commercial)

Grid and cloud databases

• The more widely used ones
  • gridgain (open source + commercial)
  • cratedb (open source)

XML databases

• The more widely used ones
  • EMC Documentum xDB (commercial)
  • eXist (open source)

Multidimensional databases

• The more widely used ones
  • Globals (commercial)
  • Intersystems Cache (commercial)

Multivalue databases

• The more widely used ones
  • OpenInsight (commercial)
  • u2 (commercial)

Event-driven databases

• The more widely used ones
  • event store (open source)
  • nevent store (open source)

Time series / streaming databases

• The more widely used ones
  • Axibase (open source)
  • kdb+ (personal edition open source)

Scientific and specialized databases

• The more widely used ones
  • bayesdb (open source)
  • gpudb (commercial)