Why are secondary indexes used in Cassandra?

Secondary indexes in Cassandra serve the purpose of enabling queries on non-primary key columns. This means that users can perform lookups on columns that are not part of the primary key, allowing for more flexible querying. While primary keys uniquely identify entries within a table and are essential for data retrieval, secondary indexes provide a means to efficiently query based on other column values without needing to restructure the data model or rely on a primary key.

By using secondary indexes, developers can retrieve data based on different criteria without having to specify the primary key or create a new table specifically for those queries. This capability is particularly useful in cases where the nature of queries may evolve, and flexibility is needed to accommodate various access patterns.

In contrast, the other choices do not accurately describe the role of secondary indexes. The option about querying on primary key columns overlooks that primary key queries do not need secondary indexes since they can be directly accessed. Merging SSTables relates to performance optimization rather than query capabilities. Lastly, data replication across multiple data centers pertains to data durability and availability, which is a fundamental aspect of Cassandra's architecture but not directly related to secondary indexes.