Apache Cassandra CDC through Sidecar

Описание: Join Bernardo Botella from Apple as he discusses the design and implementation of the next-generation Change Data Capture (CDC) feature for Apache Cassandra, primarily driven by CEP-44.

This talk, given at the Apache Cassandra Seattle Meetup, explains what CDC is and how the new architecture—leveraging the Cassandra Sidecar—solves the significant limitations of the original implementation.


What You Will Learn

What is CDC? A way to publish every database mutation as an event to a downstream service.

Use Cases: How CDC enables real-time analytics, stream processing, notifications, and efficient cache management/invalidation.

Design Principles: How the new CDC solution maintains Cassandra's core principles of scalability, consistency, high availability, and support for schema and cluster changes.

The Architecture: The Cassandra Sidecar process is the core component, responsible for retrieving and parsing commit logs, deduplicating events, ensuring quorum is met, and publishing them to the transport layer.

State Persistence: Learn how the Sidecar uses a special table within Cassandra itself to persist its state, ensuring resilience to restarts.

Supported Features: The solution supports host replacement, token moves, and cluster expansions/shrinks.


Demo: A quick demonstration of the CDC pipeline pushing inserts and deletes from Cassandra to a Kafka topic.


Speaker: Bernardo Botella, Apache Cassandra Committer, working at Apple for four years.