Netflix’s RDG query layer serves 8 B nodes via gRPC, slashing latency
The third part of Netflix’s Real‑Time Distributed Graph series reveals a gRPC‑based query API that traverses 8 billion nodes and 150 billion edges across 16‑24 async threads per instance. Level‑by‑level parallelism and volatility‑aware cache TTLs keep latency low, enabling real‑time recommendations at massive scale.