Apache Cassandra Learning Step by Step (2): Core Concepts

甩祸

　　

====15 Feb 2012, by Bright Zheng (IT进行时)====

3. Core Concepts
3.1. Keyspace
3.1.1.  Intro
A keyspace is the first dimension of the Cassandrahash, and is the container for the ColumnFamilies. Keyspaces are of roughly thesame granularity as a schema or database (i.e. a logical collection of tables)in the RDBMS world. They are the configuration and management point for columnfamilies, and is also the structure on which batch inserts are applied. In mostcases you will have one Keyspace for an application.
3.1.2. CLI

[default@unknown] drop keyspace Tutorial; 876ee520-571f-11e1-0000-242d50cf1ffd Waiting for schema agreement... ... schemas agree across the cluster [default@unknown] create keyspace Tutorial ...  with strategy_options =[{replication_factor:1}] ...  and placement_strategy ='org.apache.cassandra.locator.SimpleStrategy'; WARNING: [{}] strategy_options syntax isdeprecated, please use {} 8daea060-571f-11e1-0000-242d50cf1ffd Waiting for schema agreement... ... schemas agree across the cluster [default@unknown] use Tutorial; Authenticated to keyspace: Tutorial [default@Tutorial]

3.2. Column Family
3.2.1. Intro
A column family is a container for columns, somethinglike the TABLE in a relational system.
Model representation:

ColumnFamily
key	list
binary	1 .. * Columns

Data representation:

ColumnFamily
key	Columns
1	name	value	timestamp
	"firstname"	"Ronald"	1270073054
	"lastname"	"Mathies"	1270073054
	"birthday"	"01/01/1978"	1270073054
2	name	value	timestamp
	"firstname"	"John"	1270084021
	"lastname"	"Steward"	1270084021
	"birthday"	"01/01/1982"	1270084021

3.2.2. CLI

[default@Tutorial] drop column family StateCity; StateCity not found in current keyspace. [default@Tutorial] create column family StateCity ...         with comparator = LongType ...         and default_validation_class ='UTF8Type' ...         and key_validation_class ='UTF8Type'; d3cfa8f0-571f-11e1-0000-242d50cf1ffd Waiting for schema agreement... ... schemas agree across the cluster [default@Tutorial]

Where:
1. Comparatoris used to validate and compare/sort Column names in the CF. It has followingsupported types:

Type	Description
BytesType	Simple non-validating byte comparison (Default)
AsciiType	Similar to BytesType, but validates that input is US-ASCII
UTF8Type	UTF-8 encoded string comparison
LongType	Compares values as 64 bit longs
LexicalUUIDType	128 bit UUID compared by byte value
TimeUUIDType	Timestamp compared 128 bit version 1 UUID

Note:

• The above types are changeable subject to different versions.
  
• It is also valid to specify the fully-qualified class name to a customizedclass that extends org.apache.Cassandra.db.marshal.AbstractType.
  

3.3. Column
3.3.1. Intro
A Column consists of a name, value and a timestamp.
Model representation:

Column
name	Binary
value	Binary
timestamp	i64

Data representation:

Column
name	value	timestamp
"firstname"	"Ronald"	1270073054

3.3.2. CLI
N/A
3.4. SuperColumn
3.4.1. Intro
A SuperColumn is very similar to a ColumnFamily, itconsists of a key and a list of Columns.
Model representation:

SuperColumn
key	list
binary	1 .. * Columns

Data representation:

SuperColumn
key	Columns
1	name	value	timestamp
	"firstname"	"Ronald"	1270073054
	"lastname"	"Mathies"	1270073054
	"birthday"	"01/01/1978"	1270073054
2	name	value	timestamp
	"firstname"	"John"	1270084021
	"lastname"	"Steward"	1270084021
	"birthday"	"01/01/1982"	1270084021

The only difference to ColumnFamily is the usage. ASuperColumn is used within a ColumnFamily. So it adds an extra layer in yourdata structure, instead of having only a row which consists of a key and a listof columns. We can now have a row which consists of a key and a list of supercolumns which by itself has keys and per key a list of columns.
Once the ColumnFamily uses SuperColumn, the columntype must be “Super”. By default the column type is “Standard” which meanscommon Column.
3.4.2. CLI (TODO)
To be added here!
3.5. Others?
Please refer to http://wiki.apache.org/cassandra/APIfor more.