NOSQL Overview by luckbbs

VIEWS: 14 PAGES: 21

									     Reporter: Haiping Wang
        WAMDM Cloud Group
Mail:lulang1022@yahoo.com.cn
Outline
 Why NoSQL?
   Four trends
   History
 What is NoSQL?
   Definition
   Three fundamental theories
 NoSQL categories
 RDBMS vs. NoSQL
Trend1:data set size
 Rapid Increase of Data
   57% every year (IDC2007)
    Double every 1.5 years
   988EB (1EB=1024PB) data will be produced in 2010 (IDC)
  18 million times of all info in books
Trend2:Information connectivity
Trend3:Semi-structure
 Individualization of content!
    In the salary lists of the 1970s, all elements had exactly
     one job
    In the salary lists of the 2000s, we need 5 job columns!
     Or 8? Or 15?
 Trend accelerated by the decentralization of content
  generation that is the hallmark of the age of
  participation (“web 2.0”)
RDBMS performance
Trend4:architecture changes
NoSQL history
 The term NoSQL was first used in 1998
 Reintroduced in early 2009 by Eric Evans
 Hot in 2009
Outline
 Why NoSQL?
   Four trends
   History
 What is NoSQL?
   Definition
   Three fundamental theories
 NoSQL categories
 RDBMS vs. NoSQL
 Definition
From http://nosql-database.org/ From Wikipedia
   Original intention
                                       loosely defined class of non-
      modern web-scale databases
   Characteristics                     relational data stores
      non-relational,                    not require fixed table
      Distributed                         schemas
      open-source                        Avoid join operations
      horizontal scalable
                                          Scale horizontally
      schema-free
      easy replication support
      simple API
                                       NoSQL is NOT Only SQL
      eventually consistent / BASE
       (not ACID)
      Others…
Fundamental theories
 CAP
 BASE
    AP
 Eventual consistency
    Causal consistency
    Read-your-writes consistency
    Session consistency
    Monotonic read consistency
    Monotonic write consistency
Outline
 Why NoSQL?
   Four trends
   History
 What is NoSQL?
   Definition
   Three fundamental theories
 NoSQL categories
 RDBMS vs. NoSQL
NoSQL categories
 Key-value stores
    Based on DHTs / Amazon's Dynamo paper
    Data model: (global) collection of K-V pairs
    Example: Dynomite, Voldemort, Tokyo
 BigTable clones
    Based on Google's BigTable paper
    Data model: big table, column families
    Example: Hbase, Hypertable
NoSQL categories
 Document databases
    Inspired by Lotus Notes
    Data model: collections of K-V collections
    Example: CouchDB, MongoDB
 Graph databases
    Inspired by Euler & graph theory
    Data model: nodes, rels, K-V on both
    Example: AllegroGraph, VertexDB, Neo4j
Key-value stores
    Key            Value


                      ...
               name_€#_Stella
               mood_€#_Happy
   dog_12       birthdate%///
                  135465645)
                      …
Bigtable clones
Document databases
   Key            document

                        {
                  type: “Dog”,
                 name: “Stella”,
  dog_12        mood: “Happy”,
             birthdate: 2007-04-01
                        }
Graph databases
RDBMS vs. NoSQL
 Strong consistency vs. Eventual consistency
 Big dataset vs. HUGE datasets
 Scaling is possible vs. Scaling is easy
 SQL vs. Map-Reduce
 Good availability vs. Very high availability
Thank you!!!

								
To top