In one sentence
Amazon Keyspaces is a serverless, Cassandra-compatible wide-column database that removes the need to operate clusters, nodes, patches, and replication.
Core characteristics
- Uses CQL and common Cassandra drivers for compatible application migrations.
- Scales compute and storage with on-demand or provisioned capacity modes.
- Replicates data across multiple Availability Zones without user-managed replicas.
- Integrates with IAM, KMS, VPC endpoints, CloudTrail, and CloudWatch.
- The partition key still controls distribution; poor key design can create hot partitions.
Service boundaries
- DynamoDB: AWS-native key-value/document APIs and a broader native ecosystem, but no CQL compatibility.
- DocumentDB: MongoDB-compatible document workloads.
- RDS / Aurora: relational data, SQL, joins, and relational transactions.
- Self-managed Cassandra on EC2 / EKS: more control, but you own nodes, repair, scaling, and upgrades.
Decision cue
Choose Keyspaces when an existing Cassandra application should retain CQL and drivers while eliminating cluster operations.
Remember
Cassandra compatibility does not mean running your own Cassandra cluster. Keyspaces manages capacity and replication; the application still owns its data model and partition-key design.