Which type of data storage system is Cassandra?

A data storage system is classified partly by how it distributes its data and control across machines. A centralized system keeps the entire dataset and all control logic on a single server, so every operation passes through that one machine. A distributed system instead spreads data and control across multiple independent nodes that coordinate with each other, giving it horizontal scalability and no single point of failure.

Apache Cassandra is built exactly as a distributed, masterless (peer-to-peer) NoSQL database. Every node in a Cassandra cluster is equal and can serve both read and write requests; data is partitioned using consistent hashing and replicated across multiple nodes, so the cluster keeps working even if individual nodes fail. This matches Apache's own architecture documentation for Cassandra.

• Centralized would require Cassandra's data and control to live on one server, which contradicts its documented multi-node, masterless cluster design.

• Parallel describes simultaneous computation across processors, a processing paradigm rather than a storage-architecture classification, so it does not apply here.

• Dumb is not an established term in database or systems literature for classifying storage architectures.

Since Cassandra's own design spreads data across a cluster of cooperating nodes with no central controller, it is correctly classified as a distributed data storage system.