NoSql

NoSql ->less funcionality more performance

                Good video by Matin Fowler

                               http://www.youtube.com/watch?v=qI_g07C_Q5I

                Principles

                               Scalability

                               Performance

                               High Availability

                Colections

                               Data structured and not.

                Types of NoSql DataBases - Different Data Models

                               Key Value Store (Redis, riak)

                                               - Most simple - Simple hashmap persisted in a disk

                                               - Has a metadata

                               Document Oriented (MongoDB, Couch DB, ravenDB )

                                               - Hole mass of document structure.

                                               - Represented in form of json(can be in xml)

                                               - No schema, different from Relational

                                                               There is an implicity schema that client should know.

                                               - Always have a ID that is equivalent to a Key in the "Key value store"

                               Tabular- Column-family (BigTable(google), HBase)

                                               - combination between a row key multi-colum family for each row.

                               Graph -

                                               Basecaly a graph chart database.  Easy to move between nodes. 

                                               No actualy relationship, but structure.

                                               Arrow can be named.

                                              

                               Graph (Neo4j)

                Missing

                               No Join(to enrance scalability)

                               No Complex Transactions(change many records and commit)

                               No Constaints (implemented in app level)

                What is in

                               Query Languare(not SQL)

                               Fast Performance

                               Horizonal Scalability

                When to User

                               Store and retrieve great quantity of data

                               Storing relationships between the elements is not important

                               Dealing with growing list of elements(twitter post, blogs)

                               The data is not structured or the  structure is changing with time

                               Prototypes or fast applications need to be developed

                               Constraints and validations logic is not required in database

                Questions:

                               - Why is MySql faster than MongoDB at join operations

                                               Because mongoDB does not have any SIDE JOIN. Any join is done but client. Then it is slower than any SQL DB.

                               - Why MongoDB scales better in distributed systems

                                               Has better repplication and share better cluster

                               -