Cassandra 2.2 and 3.0 introduced several new features including support for JSON, user defined functions and aggregates, role-based authentication, and a new storage engine. Cassandra 3.0 also introduced materialized views, which allow data to be pre-computed and stored for faster queries, reducing the need for applications to manually denormalize data. The document provides examples of using new features like JSON, user defined types and functions, aggregates, and discusses how materialized views can improve performance over secondary indexes by pre-computing and storing duplicate data within Cassandra.