System design interview question
Design a Distributed Event Processing Pipeline for Real-time Analytics (like Kafka + Stream Processing)
Design a scalable distributed event processing system that ingests 5 million events per second from geographically distributed sources, processes them with sub-second latency, and powers real-time dashboards and alerts for 200M monthly active users. The system must handle diverse event schemas, guarantee at-least-once delivery semantics, support stateful stream operations (windowing, joins, aggregations), and scale horizontally across multiple data centers. Key constraints include: maintaining consistency guarantees across partitions, handling late-arriving data, supporting backpressure from slow consumers, and ensuring operational observability across thousands of nodes.
Practice this question step by step with an AI interviewer that reviews your requirements, your architecture and your trade-offs, and scores the design.