Skip to main content
In this lesson we cover Amazon Keyspaces — a managed, serverless implementation of Apache Cassandra. See the official docs: https://docs.aws.amazon.com/keyspaces/latest/devguide/. Apache Cassandra is a popular open-source NoSQL database optimized for high-throughput, low-latency reads and writes at scale. Running a production Cassandra cluster requires operational expertise: provisioning nodes, patching, capacity planning, cross‑datacenter replication, and tuning to keep the cluster healthy and performant. Amazon Keyspaces removes this operational burden by providing Cassandra compatibility without managing servers.
A presentation slide titled "AWS Keyspaces" showing an Amazon Keyspaces (for Apache Cassandra) icon on the left and four colored feature icons. The features listed are Scalable, Highly Available, Maintenance Free, and Serverless.
Clients interact with Cassandra using Cassandra Query Language (CQL), a familiar SQL-like interface for modeling and querying wide-column data. Typical self-managed Cassandra exposes an application to a cluster of peer nodes; the client communicates with nodes using CQL drivers or cqlsh.
A slide titled "Apache Cassandra" showing a client icon sending CQL requests and receiving responses from a cluster node. A ring of colored circles labeled "Node" illustrates the Cassandra cluster with arrows indicating data flow between nodes.
Keyspaces provides a fully managed, Cassandra-compatible service that handles provisioning, scaling, maintenance, and backups. It stores multiple copies of your data across Availability Zones for high availability, and integrates with AWS security and networking services so teams can focus on application logic instead of cluster operations. Capacity modes Keyspaces supports two throughput modes for reads and writes. Choose the mode that matches your traffic pattern and cost profile.
An AWS Keyspaces infographic showing the Amazon Keyspaces (for Apache Cassandra) icon at the top branching to two options. The two branches are labeled On‑Demand (pay-as-you-go, dynamic capacity) and Provisioned (capacity planning, automatic scaling) with matching icons and bullet points.
You can change a table’s capacity mode as needs evolve; however, capacity mode changes are subject to service limits (for example, switching once per day). Use On‑Demand for unpredictable workloads and Provisioned (with auto scaling) for workloads where you can forecast throughput and want to optimize cost.
Choose On-Demand for variable or unknown workloads. Choose Provisioned with auto scaling to reduce cost for predictable workloads. Remember capacity mode switches are limited (e.g., once per day), so plan migrations accordingly.
Architecture and security Clients (Cassandra drivers, cqlsh, or the Keyspaces console) connect to Keyspaces over TLS and issue CQL requests. Keyspaces persists multiple, encrypted copies of data across Availability Zones, integrates with AWS Identity and Access Management (IAM) for authentication and fine‑grained authorization, supports VPC endpoints for private connectivity, and publishes metrics to CloudWatch for monitoring and alerting. Keyspaces also supports a managed TTL feature to expire data automatically.
An architecture diagram for AWS Keyspaces (for Apache Cassandra) showing Cassandra drivers, cqlsh, and the Amazon Keyspaces console connecting (via TLS) to Amazon Keyspaces, which routes data to multiple encrypted storage backends. Side annotations note compatibility with Cassandra Query Language (CQL) and fully managed time-to-live (TTL).
Multi-region replication and active-active patterns Keyspaces can be deployed in a multi-region configuration: one Keyspaces replica per AWS Region, sharing the same table schema. Writes in one region are propagated to other regions via storage-based asynchronous replication. Replication lag is typically low (often under a second) and is designed to avoid impacting query-serving compute resources. This enables active-active global deployments where reads and writes can be performed in multiple regions to serve customers from the closest region and reduce latency. In a regional failure, traffic can be routed to other regions without losing the ability to read or write data (assuming the application is designed for multi-region failover).
A slide titled "AWS Keyspaces – Multi-Region Replication" showing a world map with three database location markers connected by dashed arrows to illustrate cross-region replication. The right side lists benefits like global reads/writes, improved business continuity, high-speed replication, and consistency/conflict resolution.
Active-active replication can create write conflicts when the same data is updated concurrently in different regions. Keyspaces includes conflict resolution mechanisms to reconcile concurrent writes, but application-level design remains critical. Consider idempotency, conflict-resolution semantics, and test concurrent-update behavior when designing globally distributed systems.
For global workloads, design and test your application for conflict resolution and idempotent writes. Understand your consistency and reconciliation patterns before moving to active-active replication.
Common use cases
  • Low-latency, high-throughput applications needing single-digit millisecond responses (telemetry ingestion, fleet and route optimization, trade monitoring).
  • Teams that require Cassandra compatibility and want to continue using Cassandra drivers and tooling.
  • Organizations that want Cassandra’s data model without operating clusters — Keyspaces handles scaling, security, backups, and maintenance.
Summary Amazon Keyspaces is a managed, serverless, Apache Cassandra-compatible database service that eliminates much of the operational overhead of running Cassandra clusters. It supports CQL, integrates with AWS security and networking, provides TTL and monitoring, and offers on-demand or provisioned capacity modes. For global, low-latency applications, Keyspaces supports multi-region active-active replication — but plan for conflict resolution and idempotent operations. Links and references

Watch Video