。

但当系统面临超高并发读取#xff08;如热点商品详情页访问#xff09;或超低延迟要求#xff08;如金融行情数据推送#xff09;时#xff0c;纯远程缓存面临两大瓶颈

在分布式系统中Redis作为分布式缓存已广泛应用。

但当系统面临超高并发读取如热点商品详情页访问或超低延迟要求如金融行情数据推送时纯远程缓存面临两大瓶颈

网络IO开销每次Redis访问需10-50ms的网络延迟带宽瓶颈单节点Redis吞吐量上限约10万QPS

通过引入Caffeine本地缓存作为一级缓存Redis作为二级缓存可实现

ms",verdana,arial,sans-serif;font-size:16px;fill:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.error-icon{fill:#552222;}#mermaid-svg-mFP5xaKAPFlnOqAn

.error-text{fill:#552222;stroke:#552222;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edge-thickness-normal{stroke-width:2px;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edge-thickness-thick{stroke-width:3.5px;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edge-pattern-solid{stroke-dasharray:0;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edge-pattern-dashed{stroke-dasharray:3;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edge-pattern-dotted{stroke-dasharray:2;}#mermaid-svg-mFP5xaKAPFlnOqAn

.marker{fill:#333333;stroke:#333333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.marker.cross{stroke:#333333;}#mermaid-svg-mFP5xaKAPFlnOqAn

ms",verdana,arial,sans-serif;font-size:16px;}#mermaid-svg-mFP5xaKAPFlnOqAn

ms",verdana,arial,sans-serif;color:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.cluster-label

text{fill:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.cluster-label

span{color:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.label

text,#mermaid-svg-mFP5xaKAPFlnOqAn

span{fill:#333;color:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.node

rect,#mermaid-svg-mFP5xaKAPFlnOqAn

.node

circle,#mermaid-svg-mFP5xaKAPFlnOqAn

.node

ellipse,#mermaid-svg-mFP5xaKAPFlnOqAn

.node

polygon,#mermaid-svg-mFP5xaKAPFlnOqAn

.node

path{fill:#ECECFF;stroke:#9370DB;stroke-width:1px;}#mermaid-svg-mFP5xaKAPFlnOqAn

.node

.label{text-align:center;}#mermaid-svg-mFP5xaKAPFlnOqAn

.node.clickable{cursor:pointer;}#mermaid-svg-mFP5xaKAPFlnOqAn

.arrowheadPath{fill:#333333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edgePath

.path{stroke:#333333;stroke-width:2.0px;}#mermaid-svg-mFP5xaKAPFlnOqAn

.flowchart-link{stroke:#333333;fill:none;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edgeLabel{background-color:#e8e8e8;text-align:center;}#mermaid-svg-mFP5xaKAPFlnOqAn

.edgeLabel

rect{opacity:0.5;background-color:#e8e8e8;fill:#e8e8e8;}#mermaid-svg-mFP5xaKAPFlnOqAn

.cluster

rect{fill:#ffffde;stroke:#aaaa33;stroke-width:1px;}#mermaid-svg-mFP5xaKAPFlnOqAn

.cluster

text{fill:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

.cluster

span{color:#333;}#mermaid-svg-mFP5xaKAPFlnOqAn

div.mermaidTooltip{position:absolute;text-align:center;max-width:200px;padding:2px;font-family:"trebuchet

ms",verdana,arial,sans-serif;font-size:12px;background:hsl(80,

100%,

#aaaa33;border-radius:2px;pointer-events:none;z-index:100;}#mermaid-svg-mFP5xaKAPFlnOqAn

:root{--mermaid-font-family:"trebuchet

未命中

redisTemplate.opsForValue().get(k);if

(obj

userMapper.selectById(userId);redisTemplate.opsForValue().set(k,

dbUser,

指标纯Redis缓存两级缓存架构提升幅度平均响应时间1-10ms100ns-1ms10-100倍数据库请求量100%1%99%Redis带宽占用100%10%-30%70%-90%

核心挑战与解决方案

问题场景集群中节点A更新数据节点B仍读旧缓存解决方案Redis

Pub/Sub

container(RedisConnectionFactory

factory)

RedisMessageListenerContainer();container.setConnectionFactory(factory);container.addMessageListener((message,

pattern)

String(message.getBody());caffeineCache.invalidate(key);

失效本地缓存},

ChannelTopic(cacheEvict));return

container;

{userMapper.updateById(user);redisTemplate.delete(key);redisTemplate.convertAndSend(cacheEvict,

key);

解决方案组合Caffeine.newBuilder().maximumSize(10_000)

TimeUnit.SECONDS)

开启监控配合Redisspring:redis:timeout:

100ms

redisTemplate.opsForValue().get(k);if

(redisVal

userMapper.selectById(userId);});

}优点完全掌控缓存逻辑

CaffeineCacheManager();manager.setCaffeine(Caffeine.newBuilder().expireAfterWrite(10,

TimeUnit.SECONDS).maximumSize(1000));return

manager;}

userMapper.selectById(userId);}

}注解对比

unlessCachePut强制更新缓存keyCacheEvict删除缓存allEntries,

beforeInvocation

Retention(RetentionPolicy.RUNTIME)

public

{Around(annotation(doubleCache))public

Object

handleCache(ProceedingJoinPoint

pjp,

pjp.proceed();updateCache(realKey,

result);

DELETE:deleteCache(realKey);return

FULLif

(caffeineCache.getIfPresent(realKey)

null)

caffeineCache.getIfPresent(realKey);if

(redisTemplate.hasKey(realKey))

{Object

redisTemplate.opsForValue().get(realKey);caffeineCache.put(realKey,

val);return

pjp.proceed();updateCache(realKey,

dbResult);return

Caffeine.newBuilder().maximumSize(10_000)

TimeUnit.SECONDS)

使用HashTag保证热点数据分片均衡user:{12345}:profile设置差异化TTL防雪崩baseTTL

random(0,

300)大Value压缩redisTemplate.setValueSerializer(new

监控指标体系

监控项健康阈值工具Caffeine命中率85%cache.stats().hitRate()Redis延迟50msRedis

Metrics

Cache注解快速实现高并发系统自定义注解Pub/Sub同步精细控制实时性要求极高Caffeine

W-TinyLFU算法98%命中率

通过两级缓存架构某电商平台在2025年大促期间成功支撑1.2亿QPSRedis成本降低60%。

正确实施该架构可让您的系统在性能和成本间获得最佳平衡