System design interview question
Design a Distributed Distributed Lock Service and Consensus System like etcd/Zookeeper at Scale
Design a highly available, strongly consistent distributed lock and coordination service that powers critical infrastructure across thousands of microservices globally. The system must handle 1M lock acquire/release operations per second with sub-100ms latency, support multi-datacenter deployment across 5+ regions with automatic failover, and provide strict linearizability guarantees for leader election, distributed transactions, and resource coordination. The service must tolerate Byzantine failures, support watch/notification mechanisms for lock state changes, and maintain data consistency across replicas even during network partitions. You should design the core consensus protocol, replication strategy, client library architecture, and operational monitoring for a system managing billions of lock acquisitions daily.
Practice this question step by step with an AI interviewer that reviews your requirements, your architecture and your trade-offs, and scores the design.