Redis on EC2 to Valkey on Kubernetes: A Zero-Downtime Case Study - Joe Heyburn, Braze

This talk presents a case study by Joe Heyburn from Braze, focusing on the migration of over 250 mission-critical Redis instances from EC2 to Kubernetes using Valkey. The session highlights the challenges faced in a 24/7 environment, where stopping traffic to copy RDB files is not feasible. Attendees will learn how Braze achieved a zero-downtime migration, enabling them to upgrade to Valkey 8.1 and reduce tail-end latency by up to 90% compared to Redis 7.2. The discussion also covers the performance gains observed, the challenges that were overcome, and key insights gained from managing stateful workloads on Kubernetes at scale.