Why cache-aside remains the default for microservices When write-through or write-behind are the right trade-offs How to stop a cache stampede: request coalescing, locks, and singleflight Why negative caching and TTL design are your best friends for noisy keys Cache invalidation strategies that preserve consistency without killing availability Actionable checklist and code snippets to implement these patterns Cache behavior decides whether a microservice scales or collapses. Implementing the right Redis caching patterns — cache-aside , write-through/write-behind , negative caching , request coalescing , and disciplined cache invalidation — turns backend storms into predictable operational pulses. The symptoms you see in production are usually familiar: sudden spikes in DB QPS and p99 latency when a hot key expires, cascading retries that double the load, or quiet churn of “not found” lookups that quietly burn CPU.…