Cassandra data modeling principles

gistfile1.txt

Design Principles
 1. Sacrifice space for time
 2. Single partition per query


Cassandra Data Modelling Process
1. Conceptual Modelling 
   Identify Entities, attributes and their relationships

   Eg: User one -> many Campaigns (relationship)
       User => name: string, email: string, location: string (attributes)

2. Define access patterns
   Find all possible queries with which data can be accessed.

   eg: 
     Q1. Find a user with specific email
     Q2. Find most recently added campaigns 

3. Logical modelling
   Logical model is the data model which support all the access patterns(queries)

   We need to do the following to create Logical model

   1. We might have create new mapping tables to support queries
   2. Identify the primary key for the table, adding partition key columns based on the required query attributes, and clustering columns in order to guarantee uniqueness and support desired sort ordering.
  
  eg: Inorder to find the users by the location we need to create a new table users_by_location.
      Attributes are listed below.
      user_id: (K)
      location: C^

      K = Partition key
      C = Clustering key

4. Physical Model
   Physical Model is created by adding CQL datatypes to the attributes


Anti patterns
  1. Full cluster scan.
  2. Full table scan. 
  3. IN queries.
  4. Queries that need a read before write. 
  5. Excess use of secondary indexes.

Reference:
http://cassandra.apache.org/doc/latest/data_modeling/index.html
https://academy.datastax.com/resources/ds220-data-modeling