当前位置：首页 > news >正文

热点Key拆分方案实现

news 来源：原创 2025/6/21 13:10:34

热点Key拆分方案实现

一、核心拆分策略

热点Key拆分的核心思想是将单个高频访问Key分解为多个子Key，分散存储到不同Redis节点，降低单节点压力。以下是具体实现方案：

二、实现方式

1. 业务层哈希分片实现

创建Key分片工具类，通过哈希取模方式分散Key：

package plus.gaga.infrastructure.redis;import org.springframework.util.StringUtils;public class KeyShardingUtil {// 分片数量，建议与Redis节点数保持一致private static final int SHARD_COUNT = 16;/*** 生成分片Key* @param originalKey 原始Key* @param shardParam 分片参数(如用户ID、商品ID等)* @return 分片后的Key*/public static String generateShardingKey(String originalKey, String shardParam) {if (StringUtils.isEmpty(originalKey) || StringUtils.isEmpty(shardParam)) {throw new IllegalArgumentException("Key and shardParam cannot be empty");}// 基于分片参数哈希取模int shardIndex = Math.abs(shardParam.hashCode()) % SHARD_COUNT;return originalKey + ":shard:" + shardIndex;}
}

2. 在服务层应用分片Key

修改，对热点活动Key进行拆分：

// ... existing code ...
import plus.gaga.infrastructure.redis.KeyShardingUtil;@Slf4j
@Service
public class LotteryStrategyServiceImpl implements LotteryStrategyService {// ... existing code ...@Overridepublic boolean tryAcquire(String activityId, String userId) {// 对热点活动ID进行分片String shardingKey = KeyShardingUtil.generateShardingKey("limit:strategy:" + activityId, userId);long timestamp = System.currentTimeMillis();String member = userId + "_" + timestamp;Long result = redisTemplate.execute(rateLimitScript,Collections.singletonList(shardingKey),1000, // 每分片QPStimestamp,member,60 // 窗口时间(秒));return result != null && result == 1;}// 热点数据查询示例@Overridepublic ActivityVO queryActivity(String activityId, String userId) {// 1. 尝试从本地缓存获取ActivityVO localActivity = localCache.get(activityId);if (localActivity != null) {return localActivity;}// 2. 从Redis分片查询String shardingKey = KeyShardingUtil.generateShardingKey("activity:" + activityId, userId);ActivityPO activityPO = (ActivityPO) redisTemplate.opsForValue().get(shardingKey);// 3. 缓存预热到本地if (activityPO != null) {localCache.put(activityId, convert(activityPO), Duration.ofMinutes(5));return convert(activityPO);}// 4. 从数据库查询并回填缓存activityPO = activityMapper.selectById(activityId);if (activityPO != null) {redisTemplate.opsForValue().set(shardingKey, activityPO, Duration.ofHours(1));localCache.put(activityId, convert(activityPO), Duration.ofMinutes(5));return convert(activityPO);}return null;}
}

3. Redis集群配置

在application.yml中配置Redis集群，确保分片Key分布到不同节点：

spring:redis:cluster:nodes:- 192.168.1.101:6379- 192.168.1.102:6379- 192.168.1.103:6379max-redirects: 3lettuce:pool:max-active: 16max-idle: 8min-idle: 4

三、高级优化策略

1. 动态分片调整

实现分片数量动态调整，应对流量变化：

// 在KeyShardingUtil中添加动态调整方法
public static void setShardCount(int count) {if (count > 0) {SHARD_COUNT = count;}
}

2. 热点检测与自动分片

集成热点Key检测，自动对超过阈值的Key进行分片：

@Component
public class HotKeyMonitor {@Autowiredprivate RedisTemplate<String, Object> redisTemplate;@Scheduled(fixedRate = 60000) // 每分钟检测一次public void monitorHotKeys() {// 获取Redis热点Key列表List<String> hotKeys = getHotKeysFromRedis();for (String key : hotKeys) {if (needSharding(key)) {// 自动分片处理shardHotKey(key);}}}
}

3. 读写分离增强

结合读写分离，将分片读请求分散到从节点：

// 配置读写分离RedisTemplate
@Bean
public RedisTemplate<String, Object> readWriteSplitRedisTemplate() {RedisTemplate<String, Object> template = new RedisTemplate<>();template.setConnectionFactory(readWriteSplitConnectionFactory());// 其他配置...return template;
}