Tuning gRPC for High Performance - Deadlines, Batching, and KeepAlives

Explore advanced gRPC performance optimization techniques in this 22-minute conference talk from KubeCon + CloudNativeCon. Dive deep into the intricacies of deadline propagation across service boundaries and understand how they impact retry behavior in distributed systems. Learn essential batching strategies, comparing manual versus automatic approaches for CPU-bound and streaming workloads to maximize throughput. Discover how keepalive pings affect long-lived connections in load-balanced environments, particularly when working with proxies like Envoy. Gain insights from real-world production tuning experiences, including handling edge cases such as connection churn, head-of-line blocking, and deadline expiration races. Master the nuanced understanding needed to fine-tune gRPC services for high-performance, low-latency systems operating under demanding real-world conditions, making this session ideal for engineers who have already deployed gRPC at scale and are ready to optimize their implementations further.