What's new in Cassandra

An overview of new features in Cassandra.

Cassandra 2.0 included major enhancements to CQL, security, and performance. CQL for Cassandra 2.0.6 adds several important features including batching of conditional updates, static columns, and increased control over slicing of clustering columns.

Key features of Cassandra 2.0 are:

• Support for lightweight transactions
  • Use of the IF keyword in CQL INSERT and UPDATE statements
  • New SERIAL consistency level
• Triggers

  The first phase of support for triggers for firing an event that executes a set of programmatic logic, which runs either inside or outside a database cluster

• CQL paging support

  Paging of result sets of SELECT statements executed over a CQL native protocol 2 connection, which eliminates the need to use the token function to page through results. For example, to page through data in this table, a simple SELECT statement after Cassandra 2.0 replaces the complex one using the token function before Cassandra 2.0.

  

• Improved authentication

  SASL support for easier and better authentication over prior versions of the CQL native protocol

• Drop column support

  Re-introduction of the ALTER TABLE DROP command

• SELECT column aliases
  Support for column aliases in a SELECT statement, similar to aliases in RDBMS SQL:

  SELECT hdate AS hired_date 
    FROM emp WHERE empid = 500

• Conditional DDL

  Conditionally tests for the existence of a table, keyspace, or index before issuing a DROP or CREATE statement using IF EXISTS or IF NOT EXISTS

• Index enhancements

  Indexing of any part, partition key or clustering columns, portion of a compound primary key

• One-off prepare and execute statements

  Use of a prepared statement, even for the single execution of a query to pass binary values for a statement, for example to avoid a conversion of a blob to a string, over a native protocol version 2 connection

• Performance enhancements
  • Off-heap partition summary
  • Eager retries support

    Sending the user request to other replicas before the query times out when a replica is unusually slow in delivering needed data

  • Compaction improvements

    Hybrid (leveled and size-tiered) compaction improvements to the leveled compaction strategy to reduce the performance overhead on read operations when compaction cannot keep pace with write-heavy workloads