Reducing gRPC Call Volume Through Caching and Batching

Explore Netflix's approach to scaling gRPC services through caching and batching techniques in this informative conference talk. Learn how the company reduces call volume by implementing client and server interceptors, automatically integrating them into services. Discover the effectiveness of on-box (Google/Guava, OHCache) and distributed (Netflix/EVcache) caches in serving over half of all Java-client gRPC traffic. Gain insights into the process of incorporating caching directives directly into proto specifications, enabling automatic cache support injection into stubs with minimal developer effort. Examine Netflix's request batching method for consolidating multiple stub invocations into a single remote call, allowing service owners to optimize call volume even when faced with suboptimal fetch patterns from consuming applications.