Denormalization refers to the process of optimizing the read performance of a database by adding redundant data or by grouping data. In Cassandra, this process is accomplished by duplicating data in multiple tables, grouping data for queries.