Pinterest implemented a generic Change Data Capture (CDC) solution to address inconsistencies in existing, isolated solutions. This new system utilizes Red Hat Debezium and is designed for reliability, scalability, and low latency. The architecture separates the control plane, managing system state and configuration, from the data plane, processing changes and sending them to Kafka. Kafka stores the CDC data, which users can then consume. The implementation overcame several challenges, including scalability issues, rebalancing timeouts, and duplicate tasks. Solutions involved bootstrapping, rate limiting, adjusting timeout configurations, and upgrading Kafka. The improvements resulted in stable system performance and significantly reduced failover recovery time. Future plans include enhancing scalability, implementing disaster recovery using CDC, and creating a near real-time database ingestion system. The project's success involved contributions from multiple Pinterest teams. Finally, the blog post includes a disclaimer regarding trademarks.