Assess students' ability to apply the Raft consensus algorithm to design and analyze a fault-tolerant replicated key-value store. Tasks should include specifying system architecture (leader election, log replication, membership changes), state machine semantics and client request handling to achieve linearizability, mechanisms for persistence, snapshotting and log compaction, procedures for leader crashes, network partitions and node recovery, and a quantitative evaluation of availability, consistency, and recovery time under fault scenarios. Require justification of design choices and trade-offs.