Where is my cache? Architectural patterns for caching microservices by example
summary
Compares five concrete deployment patterns and their tradeoffs: Embedded distributed with Hazelcast saves network hops but ties data to replicas and JVMs; Client‑server (Redis/Memcached or Hazelcast Cloud) centralizes storage for independent scaling but requires ops and same cloud/region/VPC peering for low latency; Kubernetes sidecar gives local low‑latency isolation per node; Reverse proxy (NGINX) caching is protocol‑level, disk‑backed and non‑distributed; Proxy sidecar injects caching without code changes but makes invalidation harder. Demo and Helm charts for Hazelcast exist; proof‑of‑concept reverse‑proxy sidecar works.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:55About Hazelcast
- 2:23Why cache
- 3:57Embedded cache
- 11:55Clientserver
- 15:33Cloud
- 19:18Pros Cons
- 21:40Sidecar
- 29:52The Good
- 32:55The Bad
- 35:49Summary
- 38:06Questions
related talks
47mArchitectural Patterns for Caching Microservices
34mHigh Performance Cloud-Native Microservices
53mLeveraging Caching For Cloud Native Applications
34mAdvanced Caching Patterns used by 2000 microservices
42m3 easy performance improvements in your microservices architecture
05mStartup Showcase: Hazelcast
30mCaching patterns for lazy devs for lazy loading
50mCaching with Spring: Advanced Topics and Best Practices