Spring Boot + Caffeine Cache Demo

A tiny, production-style starter showing Spring Cache abstraction with Caffeine.

What’s inside

• Spring Boot 3.3.x, Java 21
• @EnableCaching + Caffeine spec via application.yml
• REST endpoints to fetch/update/delete users
• Single-flight caching (sync=true) to avoid stampede
• Actuator with cache metrics exposed

Run

Prereqs: Java 21 and Maven 3.9+

cd caffeine-cache-demo
mvn spring-boot:run

Try it

First call is slow (~400ms), subsequent calls are cached and fast.

# First call (cache miss -> slow)
curl -s localhost:8080/api/users/1 | jq

# Second call (cache hit -> fast)
curl -s localhost:8080/api/users/1 | jq

# See how many times the underlying load ran
curl -s localhost:8080/api/debug/loadCount
# -> 1

# Update (write-through into cache)
curl -s -X PUT "localhost:8080/api/users/1?name=Vinod" | jq

# Evict by delete
curl -i -X DELETE localhost:8080/api/users/1

# Miss again (load count increments)
curl -s localhost:8080/api/users/1 | jq
curl -s localhost:8080/api/debug/loadCount

Configuration

Edit spring.cache.caffeine.spec in application.yml:

maximumSize=10000,expireAfterWrite=10m,recordStats

Common options: expireAfterAccess, initialCapacity, refreshAfterWrite (use with loading cache patterns).

Metrics

With Actuator:

• GET /actuator/metrics/cache.gets (tags cache=users etc.)
• GET /actuator/metrics/cache.hit.ratio

You can wire Prometheus/Grafana later; Micrometer auto-binds Caffeine stats.