Use Valkey — the Apache-2.0 fork of Redis maintained by the Linux Foundation — as the cache, stream, and coordination primitive at the heart of a Rust service. This course is the most direct path from `cargo add redis` to a real-time data pipeline that survives partitions and operator changes.

You begin with the license rift and what wire compatibility means in practice, then dive into the four data shapes that drive every Valkey workload — strings with TTL, hashes with per-field expiry, lists as queues, and sorted sets as the leaderboard primitive. From there the course turns to streams + consumer groups (the data-engineering core), XADD and XREADGROUP for exactly-once delivery, plus XPENDING + XCLAIM for failure handling. The back half covers the patterns that make Valkey production-grade in Rust: pipelining for round-trip math, MULTI/EXEC with WATCH for multi-key atomicity, EVAL and the script cache for server-side atomicity, connection-pool sizing under load, and criterion for honest p50/p99 benchmarking. The course closes with the operational story — RDB vs AOF, replication, Sentinel failover, cluster sharding, and the split-brain accounting that "highly available" actually buys you.