Cassandra Architecture

Apache Cassandra is not another relational database in market. Instead of using relational model, it uses key-value map to store its data. The structure is more or less can be explain as in the following picture:



Cluster: Cassandra is designed to be distributed over several nodes/machines. A cluster consists of several nodes. I've only ever used Cassandra in a single node which is my computer.

Keyspace: A cluster consists of several keyspaces. Keyspace is the place where our data reside. A keyspace could have several Column Family or Super Column Family.

Column Family and Super Column Family: Both Column Family and Super Column Family is a collection of rows, just like a table is a collection of rows in relational database.

Row: A row consists of columns; key-value columns for a row in Column Family, or Super Columns for a row in Super Column Family.

Super Column: It is sort of container of sub-columns (which are of type Key-value Column).

Key-value Column: The most basic data structure in Cassandra where the actual data is saved as byte. The behavior is a lot like Java Hash data type.

Here are examples of Cassandra data retention and retrieval:

COLUMN_FAMILY_NAME[ROW_KEY_NAME][COLUMN]

User['user1']['name'] = 'Rochmat Santoso';
User['user1']['address'] = 'Surabaya';
User['user2']['name'] = 'Barbie';
User['user2']['address'] = 'Malibu';

SUPER_COLUMN_FAMILY_NAME[ROW_KEY_NAME][SUPER_COLUMN_NAME][COLUMN]

Province['province1']['info']['name'] = 'East Java';
Province['province1']['info']['capital'] = 'Surabaya';
Province['province1']['city']['sby'] = 'Surabaya';
Province['province1']['city']['mlg'] = 'Malang';
Province['province1']['city']['sda'] = 'Sidoarjo';

Province['province2']['info']['name'] = 'West Java';
Province['province2']['info']['capital'] = 'Bandung';
Province['province2']['city']['bdg'] = 'Bandung';
Province['province2']['city']['tgr'] = 'Tangerang';