缓存工具CacheUtil - 并发环境的缓存值存取

时间 2019-12-14

标签缓存工具 cacheutil 并发环境存取繁體版

原文原文链接

缓存工具CacheUtil - 并发环境的缓存值存取

目的

适合并发环境的缓存值存取
读取缓存值时，只需关注数据来源。不用再关注将源数据存入缓存等后续处理。
应用程序N次读取数据时，数据源读取一次，缓存读取N-1次。

设计

当从缓存查找失败，则去数据源获取。获取成功，存入缓存并返回。java

实现

Sourcable
java /** * 可溯源的。可从数据源获取数据 */ public interface Sourcable { /** * 从数据源获取 */ <T> T get(); }spring

CacheUtil
```java
public class CacheUtil {
private static final transient Log logger = LogFactory.getLog(CacheUtil.class);shell

// @Resource(name = "memCacheServiceImpl")
private static CacheService cache;数据库

/**
   * 获取value
   * 若是从缓存查找失败，则尝试从数据源获取，获取成功，存入缓存并返回。
   * @param key
   * @param source Sourcable的实现，用于读取源数据。
   * @param isShort 是否(短期)暂存
   */
  public static <T> T get(String key, Sourcable source, boolean isShort) {
      logger.debug("-> [" + Thread.currentThread().getId() + "] Enter");
      T value = get(key);
      if (value == null && source != null) {
          // 缓存查找失败，尝试从数据源获取
          logger.debug("-> [" + Thread.currentThread().getId() + "] try to Lock");
          key = key.intern();
          synchronized (key) {
              logger.debug("-> [" + Thread.currentThread().getId() + "] Locked");
              value = get(key);
              // 双重检查。防止多线程重复从数据源读取
              if (value == null) {
                  // 从数据源获取
                  value = source.get();
                  if (value != null) {
                      // 存入缓存
                      if (isShort) {
                          // 暂存cache
                          put4short(key, value);
                      } else {
                          // 常规存入cache，正常有效期
                          put(key, value);
                      }
                  }
                  logger.debug("-> [" + Thread.currentThread().getId() + "] Loaded");
              }
          }
      }
      logger.debug("-> [" + Thread.currentThread().getId() + "] get: " + value);
      return value;
  }

  // 其余代码省略..

}
```缓存

测试

CacheUtilTest
```java
public class CacheUtilTest extends SpringTest {多线程

@Test
  public void get() throws Exception {
      final Long resourceId = 173L;
      final String cacheKey = Resource.class.getName() + '#' + resourceId;

      // 清除
      CacheUtil.delete(cacheKey);

      // 1000个线程并发存取
      int num = 1000;
      Executor executor = Executors.newFixedThreadPool(num);
      final CountDownLatch latch = new CountDownLatch(num);
      for (int i = 0; i < num; i++) {
          executor.execute(() -> {
              // new String(cacheKey)。 用于验证不一样key对象时，同步锁的有效性
              String key = new String(cacheKey);
              // 读取缓存值时，只需关注数据来源。不用再关注将源数据存入缓存。
              Resource resource = CacheUtil.get(key, new Sourcable() {
                      @Override
                      public <T> T get() {
                          List<Resource> resources = DAOUtil.findByHql("FROM Resource WHERE resourceId=" + resourceId);
                          if (CollectionUtils.isNotEmpty(resources)) {
                              return resources.get(0);
                          }
                          return null;
                      }
                  });
              latch.countDown();
          });
      }
      latch.await();
  }

}
```并发

附录

SpringTest
java @RunWith(SpringJUnit4ClassRunner.class) @ContextConfiguration("classpath:context/spring.xml") public abstract class SpringTest {}ide
结果
10个线程运行结果：

[PAY][DEBUG] 2016-10-14 15:14:20,526 (CacheUtil.java:46).get - [pool-1-thread-2] 14030 
-> [29] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,527 (CacheUtil.java:46).get - [pool-1-thread-8] 14031 
-> [35] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,527 (CacheUtil.java:46).get - [pool-1-thread-10] 14031 
-> [37] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,528 (CacheUtil.java:46).get - [pool-1-thread-9] 14032 
-> [36] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,527 (CacheUtil.java:46).get - [pool-1-thread-4] 14031 
-> [31] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,527 (CacheUtil.java:46).get - [pool-1-thread-7] 14031 
-> [34] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,526 (CacheUtil.java:46).get - [pool-1-thread-6] 14030 
-> [33] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,526 (CacheUtil.java:46).get - [pool-1-thread-5] 14030 
-> [32] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,555 (CacheUtil.java:49).get - [pool-1-thread-7] 14059 
-> [34] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,557 (CacheUtil.java:53).get - [pool-1-thread-7] 14061 
-> [34] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,561 (CacheUtil.java:49).get - [pool-1-thread-5] 14065 
-> [32] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,555 (CacheUtil.java:49).get - [pool-1-thread-4] 14059 
-> [31] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,547 (CacheUtil.java:49).get - [pool-1-thread-10] 14051 
-> [37] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,547 (CacheUtil.java:49).get - [pool-1-thread-8] 14051 
-> [35] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,547 (CacheUtil.java:49).get - [pool-1-thread-9] 14051 
-> [36] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,547 (CacheUtil.java:49).get - [pool-1-thread-2] 14051 
-> [29] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,528 (CacheUtil.java:46).get - [pool-1-thread-1] 14032 
-> [28] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,527 (CacheUtil.java:46).get - [pool-1-thread-3] 14031 
-> [30] Enter
[PAY][DEBUG] 2016-10-14 15:14:20,565 (CacheUtil.java:49).get - [pool-1-thread-6] 14069 
-> [33] try to Lock
Hibernate: FROM Resource WHERE resourceId=173
[PAY][DEBUG] 2016-10-14 15:14:20,595 (CacheUtil.java:49).get - [pool-1-thread-1] 14099 
-> [28] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,599 (CacheUtil.java:49).get - [pool-1-thread-3] 14103 
-> [30] try to Lock
[PAY][DEBUG] 2016-10-14 15:14:20,639 (CacheUtil.java:72).get - [pool-1-thread-7] 14143 
-> [34] Loaded
[PAY][DEBUG] 2016-10-14 15:14:20,639 (CacheUtil.java:53).get - [pool-1-thread-3] 14143 
-> [30] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,639 (CacheUtil.java:76).get - [pool-1-thread-7] 14143 
-> [34] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,645 (CacheUtil.java:53).get - [pool-1-thread-1] 14149 
-> [28] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,645 (CacheUtil.java:76).get - [pool-1-thread-3] 14149 
-> [30] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,651 (CacheUtil.java:76).get - [pool-1-thread-1] 14155 
-> [28] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,651 (CacheUtil.java:53).get - [pool-1-thread-6] 14155 
-> [33] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,658 (CacheUtil.java:76).get - [pool-1-thread-6] 14162 
-> [33] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,658 (CacheUtil.java:53).get - [pool-1-thread-2] 14162 
-> [29] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,665 (CacheUtil.java:76).get - [pool-1-thread-2] 14169 
-> [29] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,665 (CacheUtil.java:53).get - [pool-1-thread-9] 14169 
-> [36] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,671 (CacheUtil.java:76).get - [pool-1-thread-9] 14175 
-> [36] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,671 (CacheUtil.java:53).get - [pool-1-thread-8] 14175 
-> [35] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,676 (CacheUtil.java:76).get - [pool-1-thread-8] 14180 
-> [35] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,676 (CacheUtil.java:53).get - [pool-1-thread-10] 14180 
-> [37] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,681 (CacheUtil.java:76).get - [pool-1-thread-10] 14185 
-> [37] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,681 (CacheUtil.java:53).get - [pool-1-thread-4] 14185 
-> [31] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,686 (CacheUtil.java:76).get - [pool-1-thread-4] 14190 
-> [31] get: /uploadfiles/resource/2014122313575119.png
[PAY][DEBUG] 2016-10-14 15:14:20,686 (CacheUtil.java:53).get - [pool-1-thread-5] 14190 
-> [32] Locked
[PAY][DEBUG] 2016-10-14 15:14:20,694 (CacheUtil.java:76).get - [pool-1-thread-5] 14198 
-> [32] get: /uploadfiles/resource/2014122313575119.png

只有一个id为34的线程Loaded，其他线程均等待线程34 处理（从数据库查询结果并存入缓存）完成后，再查找缓存从而命中数据。工具

问题

数据源失效问题当数据源不存在数据时，一直缓存失败，则会一直从数据源获取数据。从而形成数据源负担。为防止这个状况，能够再进一步作个过滤保障机制。