Spring Caching

Spring cache abstraction

Spring provides its own cache solution in 3.x. And it defines a CacheManager SPI for third party cache solutions. By default, in the Spring core, it provides two built-in implementations, which based on JDK ConcurrentMap, Ehcache, Guava, etc.

ConcurrentMap is an in-memory cache solution, it is useful in the development stage.

Enable Spring Caching

Add @EnableCaching(mode=AdviceMode.ASPECTJ) annotation on the configuration class if you are using Java based configuration.

@Configuration
...
@EnableCaching(mode=AdviceMode.ASPECTJ)
public class JpaConfig {...}

Then, specify a CacheManager provider in your configuration.

@Override
@Bean
public CacheManager cacheManager() {
 return new ConcurrentMapCacheManager();
}

@Override
@Bean
public KeyGenerator keyGenerator() {
 return new SimpleKeyGenerator();
}

Code sample

Spring provides some annotations for applying the cache in your classes, @Cacheable, @CachePut, @CacheEvict, the former two are use for creating cache, and the last is for remove caches. @CachePut always replace the current cache value even there is no change.

@Repository
public class JpaConferenceDaoImpl implements ConferenceDao {

 @Override
 @Cacheable(value="conference", key="#id")
 public Conference findById(Long id) {
 ...
 }

 @Override
 @CacheEvict(value="conference", allEntries=true)
 public void deleteAll() {
 ...
 }

 @Override
 @CachePut(value="conference", key="#result.id")
 public Conference findBySlug(String slug) {
 ...
 }
}

Test it now.

@Test
@Transactional
public void retrieveConference() {

 Long id = conferenceDao.save(newConference());

 entityManager.flush();
 assertTrue(id != null);
 log.debug("id @=" + id);
 Conference conference = conferenceDao.findById(id);

 assertTrue(conference != null);

 assertTrue("JUD2013".equals(conference.getName()));

 // query by slug
 conference = conferenceDao.findBySlug("jud-2013");

 assertTrue(conference !=  null);

 assertTrue("JUD2013".equals(conference.getName()));

 log.debug("@@@@@@@@@@@@@@loading from cache@@@@@@@@@@@@@@@@@@@@");

 Conference conference2 = conferenceDao.findById(conference.getId());
 assertTrue(conference2 != null);
}

The last query operation does not hint database at all, it is loaded from cache.

Summary

Compare to the JPA 2nd level cache, the Spring solutions has some advantages.

• It is more flexible.

You can define which methods you create and evict caches. In JPA, the cache is created when the object is persisted, you can change the strategy. As the above example codes, creating cache happens when fetching a conference from database, and do nothing when saving a conference into database.

• Spring cache can be used in any data persistence solutions. As the above example codes, we are using cache on methods, it does not depend on JPA at all, so cache can be used in Jdbc and other data persistence solution.

Besides the built-in cache implementations, Spring Data project provides CacheManager for GemFire and Redis. And JBoss infinispan(Maven artifact is infinispan-spring) also provides an CacheManager implementation.

The example codes are hosted on my Github account.

https://github.com/hantsy/spring-sandbox4