NoSQL databases

NoSQL database types

Key-value stores: simplest, item stored in the form of key/value pairs. Each Key is unique and accepts only strings, whereas the value corresponding to the particular Key can accept String, JSON, XML, etc.
- Use Case: Session info, user preference, shopping carts, produce recommendations, etc
- Products: Redis, Voldemort, Riak, and Amazon’s Dynamo
Document stores: extension of key-value store, pair each key with a complex data structure known as a document (i.e. JSON, XML, etc)
- Use Case: User profiles, Management of content, etc
- Products: MongoDB and CouchDB
Column-oriented database: store columns of data together instead of rows and are optimized for queries over large datasets
- Use Case: User Preferences, Business Intelligence, Managing data warehouses, etc
- Products: Cassandra, HBase, Hypertable, Google's BigTable
Graph database: store information about networks, such as social connections
- Use Case: Social Networking, Recommendation Engine, Logistics, Risk Assessment, Fraud detection, etc
- Products: Neo4J and HyperGraphDB

Cassandra is de-centralized, which means all nodes are the same (its called server-symmetry). There is no leader-follower structure, all the nodes follows P2P (peer to peer) gossip protocol, so its highly available
Cassandra is highly scalable, new node will automatically be discovered and there is no reboot needed

Difference	Cassandra	MongoDB
DB structure	Unstructured data	JSON-like documents
Index	Primary Key, but no secondary indexes supported	Index, if no index then each file is searched which could slow down read times
Query	CQL (like SQL)	More like programming
Replication	leader-leader	leader-follower with auto-election