TLDR Tech / DevOps
Building Service Topology at Scale: Architecture, Challenges, and Lessons Learned
Netflix built a real-time service dependency mapping system that processes millions of network flow records per second across multiple regions, solving massive scaling challenges including "hot node" issues where some instances handled 100x more traffic than others and garbage collection pauses that consumed more CPU than business logic. The team deployed a three-stage distributed pipeline using streaming architecture with backpressure handling, switched from gRPC to Server-Sent Events for better performance, and added time-travel query capabilities that let engineers reconstruct topology from any point in time for incident investigation.
Read the full story