Keyspaces

Keyspaces are a fundamental organizing construct in Cassandra and some other distributed databases. A keyspace is the top‑level namespace that contains tables and defines how data is replicated and stored across the cluster. It serves as the unit of data distribution, durability, and logical separation within a cluster.

Within a keyspace you create tables (also called column families). Each keyspace has properties that govern its behavior, notably the replication strategy, replication factor, and durable_writes. The replication strategy specifies how data is replicated. SimpleStrategy replicates data across the cluster using a single replication factor and is suited for single-datacenter deployments. NetworkTopologyStrategy allows different replication factors per data center, enabling multi-datacenter deployment patterns. The replication_factor sets the number of replicas for each data partition. Durable_writes controls whether writes are durably recorded; when true (the default) each write is stored in the commit log and memtable before returning success, while false may improve performance but risks data loss on a crash.

Keyspaces are named namespaces; within a keyspace the tables store rows. The keyspace’s replication policy affects where copies of data exist, while the partition key of a table determines which node stores a given row. Management of keyspaces is typically done with CQL commands: CREATE KEYSPACE, ALTER KEYSPACE, and DROP KEYSPACE, including options for REPLICATION and DURABLE_WRITES. Changes to replication factor affect new writes, while existing data may require repair (such as nodetool repair) to balance across nodes after topology changes.