Redis中的某一热点数据缓存过期了，此时有大量请求访问怎么办?

1、提前设置热点数据永不过期

2、分布式中用redis分布式锁（锁可以在多个 JVM 实例之间协调）、单体中用synchronized（锁只在同一个 JVM 内有效）

编写服务类

import com.redisson.api.RLock;
import com.redisson.api.RedissonClient;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.data.redis.core.StringRedisTemplate;
import org.springframework.stereotype.Service;import java.util.concurrent.TimeUnit;@Service
public class CacheService {@Autowiredprivate StringRedisTemplate redisTemplate;@Autowiredprivate RedissonClient redissonClient;private static final String HOT_DATA_KEY = "hotData";private static final String LOCK_KEY = "hotDataLock";public String getHotData() {// 尝试从 Redis 中获取热点数据String hotData = redisTemplate.opsForValue().get(HOT_DATA_KEY);if (hotData == null) {// 获取分布式锁RLock lock = redissonClient.getLock(LOCK_KEY);try {// 尝试加锁，最多等待100ms，锁的过期时间为30秒if (lock.tryLock(100, 30, TimeUnit.SECONDS)) {try {// 再次检查缓存是否过期（双重检查）hotData = redisTemplate.opsForValue().get(HOT_DATA_KEY);if (hotData == null) {// 缓存确实过期，从数据库加载数据hotData = loadHotDataFromDatabase();// 将数据存入 Redis，设置过期时间为10分钟redisTemplate.opsForValue().set(HOT_DATA_KEY, hotData, 10, TimeUnit.MINUTES);}} finally {// 释放锁lock.unlock();}}} catch (InterruptedException e) {Thread.currentThread().interrupt();}}return hotData;}private String loadHotDataFromDatabase() {// 模拟从数据库加载数据return "Hot Data from Database";}
}

模拟多个请求

import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.boot.CommandLineRunner;
import org.springframework.stereotype.Component;import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;@Component
public class CacheTestRunner implements CommandLineRunner {@Autowiredprivate CacheService cacheService;@Overridepublic void run(String... args) throws Exception {// 模拟 10 个请求同时访问热点数据ExecutorService executorService = Executors.newFixedThreadPool(10);for (int i = 0; i < 10; i++) {executorService.submit(() -> {String hotData = cacheService.getHotData();System.out.println("Thread " + Thread.currentThread().getId() + " got hot data: " + hotData);});}executorService.shutdown();}
}

Thread 12 got hot data: Hot Data from Database
Thread 13 got hot data: Hot Data from Database
Thread 14 got hot data: Hot Data from Database
...

所有线程最终都会获取到相同的数据，但只有第一个线程会去加载数据，避免了缓存击穿问题。

单体应用

在单体应用中，所有请求都运行在同一个 JVM 实例中，因此可以使用 synchronized 来同步线程。

java复制

@Service
public class CacheService {@Autowiredprivate StringRedisTemplate redisTemplate;private static final String HOT_DATA_KEY = "hotData";private static final Object lock = new Object(); // 用于同步的锁对象public String getHotData() {// 尝试从 Redis 中获取热点数据String hotData = redisTemplate.opsForValue().get(HOT_DATA_KEY);if (hotData == null) {synchronized (lock) { // 使用 synchronized 同步// 再次检查缓存是否过期（双重检查）hotData = redisTemplate.opsForValue().get(HOT_DATA_KEY);if (hotData == null) {// 缓存确实过期，从数据库加载数据hotData = loadHotDataFromDatabase();// 将数据存入 Redis，设置过期时间为10分钟redisTemplate.opsForValue().set(HOT_DATA_KEY, hotData, 10, TimeUnit.MINUTES);}}}return hotData;}private String loadHotDataFromDatabase() {// 模拟从数据库加载数据return "Hot Data from Database";}
}