DataStax Serverless: What We Did and Why It’s a Game Changer

Apache Cassandra is a highly scalable database known for its write latency and throughput. However, it faces challenges when scaling down due to the need to scale both compute and storage simultaneously. To address this issue, DataStax developed Astra, a serverless version of Cassandra that separates compute and storage, allowing users to pay only for operations they consume and data stored or transmitted. This approach leads to significant cost savings and improved efficiency. Astra also supports external JWT token providers, simplifying user management, authentication, and authorization logic in applications. The serverless version of Cassandra is built on Kubernetes, which provides native integrations for dynamic scalability and leverages etcd for consistent metadata across nodes. DataStax plans to continue optimizing Astra's performance, scalability, and cost-effectiveness while proposing the changes made into the Apache Cassandra project as part of the CEP process.