31. Caching

The Spring Framework provides support for transparently adding caching to an application. At its core, the abstraction applies caching to methods, thus reducing the number of executions based on the information available in the cache. The caching logic is applied transparently, without any interference to the invoker. Spring Boot auto-configures the cache infrastructure as long as caching support is enabled via the @EnableCaching annotation.

[Note]Note

Check the relevant section of the Spring Framework reference for more details.

In a nutshell, adding caching to an operation of your service is as easy as adding the relevant annotation to its method, as shown in the following example:

   import org.springframework.cache.annotation.Cacheable
import org.springframework.stereotype.Component;

@Component
public class MathService {

	@Cacheable("piDecimals")
	public int computePiDecimal(int i) {
		// ...
	}

}

This example demonstrates the use of caching on a potentially costly operation. Before invoking computePiDecimal, the abstraction looks for an entry in the piDecimals cache that matches the i argument. If an entry is found, the content in the cache is immediately returned to the caller, and the method is not invoked. Otherwise, the method is invoked, and the cache is updated before returning the value.

[Caution]Caution

You can also use the standard JSR-107 (JCache) annotations (such as @CacheResult) transparently. However, we strongly advise you to not mix and match the Spring Cache and JCache annotations.

If you do not add any specific cache library, Spring Boot auto-configures a simple provider that uses concurrent maps in memory. When a cache is required (such as piDecimals in the preceding example), this provider creates it for you. The simple provider is not really recommended for production usage, but it is great for getting started and making sure that you understand the features. When you have made up your mind about the cache provider to use, please make sure to read its documentation to figure out how to configure the caches that your application uses. Nearly all providers require you to explicitly configure every cache that you use in the application. Some offer a way to customize the default caches defined by the spring.cache.cache-names property.

[Tip]Tip

It is also possible to transparently update or evict data from the cache.

[Note]Note

If you use the cache infrastructure with beans that are not interface-based, make sure to enable the proxyTargetClass attribute of @EnableCaching.

31.1 Supported Cache Providers

The cache abstraction does not provide an actual store and relies on abstraction materialized by the org.springframework.cache.Cache and org.springframework.cache.CacheManager interfaces.

If you have not defined a bean of type CacheManager or a CacheResolver named cacheResolver (see CachingConfigurer), Spring Boot tries to detect the following providers (in the indicated order):

[Tip]Tip

It is also possible to force a particular cache provider by setting the spring.cache.type property. Use this property if you need to disable caching altogether in certain environment (such as tests).

[Tip]Tip

Use the spring-boot-starter-cache “Starter” to quickly add basic caching dependencies. The starter brings in spring-context-support. If you add dependencies manually, you must include spring-context-support in order to use the JCache, EhCache 2.x, or Guava support.

If the CacheManager is auto-configured by Spring Boot, you can further tune its configuration before it is fully initialized by exposing a bean that implements the CacheManagerCustomizer interface. The following example sets a flag to say that null values should be passed down to the underlying map:

@Bean
public CacheManagerCustomizer<ConcurrentMapCacheManager> cacheManagerCustomizer() {
	return new CacheManagerCustomizer<ConcurrentMapCacheManager>() {
		@Override
		public void customize(ConcurrentMapCacheManager cacheManager) {
			cacheManager.setAllowNullValues(false);
		}
	};
}
[Note]Note

In the preceding example, an auto-configured ConcurrentMapCacheManager is expected. If that is not the case (either you provided your own config or a different cache provider was auto-configured), the customizer is not invoked at all. You can have as many customizers as you want, and you can also order them by using @Order or Ordered.

31.1.1 Generic

Generic caching is used if the context defines at least one org.springframework.cache.Cache bean. A CacheManager wrapping all beans of that type is created.

31.1.2 JCache (JSR-107)

JCache is bootstrapped through the presence of a javax.cache.spi.CachingProvider on the classpath (that is, a JSR-107 compliant caching library exists on the classpath), and the JCacheCacheManager is provided by the spring-boot-starter-cache “Starter”. Various compliant libraries are available, and Spring Boot provides dependency management for Ehcache 3, Hazelcast, and Infinispan. Any other compliant library can be added as well.

It might happen that more than one provider is present, in which case the provider must be explicitly specified. Even if the JSR-107 standard does not enforce a standardized way to define the location of the configuration file, Spring Boot does its best to accommodate setting a cache with implementation details, as shown in the following example:

   # Only necessary if more than one provider is present
spring.cache.jcache.provider=com.acme.MyCachingProvider
spring.cache.jcache.config=classpath:acme.xml
[Note]Note

When a cache library offers both a native implementation and JSR-107 support, Spring Boot prefers the JSR-107 support, so that the same features are available if you switch to a different JSR-107 implementation.

[Tip]Tip

Spring Boot has general support for Hazelcast. If a single HazelcastInstance is available, it is automatically reused for the CacheManager as well, unless the spring.cache.jcache.config property is specified.

There are two ways to customize the underlying javax.cache.cacheManager:

  • Caches can be created on startup by setting the spring.cache.cache-names property. If a custom javax.cache.configuration.Configuration bean is defined, it is used to customize them.
  • org.springframework.boot.autoconfigure.cache.JCacheManagerCustomizer beans are invoked with the reference of the CacheManager for full customization.
[Tip]Tip

If a standard javax.cache.CacheManager bean is defined, it is wrapped automatically in an org.springframework.cache.CacheManager implementation that the abstraction expects. No further customization is applied to it.

31.1.3 EhCache 2.x

EhCache 2.x is used if a file named ehcache.xml can be found at the root of the classpath. If EhCache 2.x is found, the EhCacheCacheManager provided by the spring-boot-starter-cache “Starter” is used to bootstrap the cache manager. An alternate configuration file can be provided as well, as shown in the following example:

spring.cache.ehcache.config=classpath:config/another-config.xml

31.1.4 Hazelcast

Spring Boot has general support for Hazelcast. If a HazelcastInstance has been auto-configured, it is automatically wrapped in a CacheManager.

31.1.5 Infinispan

Infinispan has no default configuration file location, so it must be specified explicitly. Otherwise, the default bootstrap is used.

spring.cache.infinispan.config=infinispan.xml

Caches can be created on startup by setting the spring.cache.cache-names property. If a custom ConfigurationBuilder bean is defined, it is used to customize the caches.

[Note]Note

The support of Infinispan in Spring Boot is restricted to the embedded mode and is quite basic. If you want more options, you should use the official Infinispan Spring Boot starter instead. See Infinispan’s documentation for more details.

31.1.6 Couchbase

If the Couchbase Java client and the couchbase-spring-cache implementation are available and Couchbase is configured, a CouchbaseCacheManager is auto-configured. It is also possible to create additional caches on startup by setting the spring.cache.cache-names property. These caches operate on the Bucket that was auto-configured. You can also create additional caches on another Bucket by using the customizer. Assume you need two caches (cache1 and cache2) on the "main" Bucket and one (cache3) cache with a custom time to live of 2 seconds on the “another” Bucket. You can create the first two caches through configuration, as follows:

spring.cache.cache-names=cache1,cache2

Then you can define a @Configuration class to configure the extra Bucket and the cache3 cache, as follows:

@Configuration
public class CouchbaseCacheConfiguration {

	private final Cluster cluster;

	public CouchbaseCacheConfiguration(Cluster cluster) {
		this.cluster = cluster;
	}

	@Bean
	public Bucket anotherBucket() {
		return this.cluster.openBucket("another", "secret");
	}

	@Bean
	public CacheManagerCustomizer<CouchbaseCacheManager> cacheManagerCustomizer() {
		return c -> {
			c.prepareCache("cache3", CacheBuilder.newInstance(anotherBucket())
					.withExpiration(2));
		};
	}

}

This sample configuration reuses the Cluster that was created through auto-configuration.

31.1.7 Redis

If Redis is available and configured, a RedisCacheManager is auto-configured. It is possible to create additional caches on startup by setting the spring.cache.cache-names property and cache defaults can be configured by using spring.cache.redis.* properties. For instance, the following configuration creates cache1 and cache2 caches with a time to live of 10 minutes:

spring.cache.cache-names=cache1,cache2
spring.cache.redis.time-to-live=600000
[Note]Note

By default, a key prefix is added so that, if two separate caches use the same key, Redis does not have overlapping keys and cannot return invalid values. We strongly recommend keeping this setting enabled if you create your own RedisCacheManager.

[Tip]Tip

You can take full control of the configuration by adding a RedisCacheConfiguration @Bean of your own. This can be useful if you’re looking for customizing the serialization strategy.

31.1.8 Caffeine

Caffeine is a Java 8 rewrite of Guava’s cache that supersedes support for Guava. If Caffeine is present, a CaffeineCacheManager (provided by the spring-boot-starter-cache “Starter”) is auto-configured. Caches can be created on startup by setting the spring.cache.cache-names property and can be customized by one of the following (in the indicated order):

  1. A cache spec defined by spring.cache.caffeine.spec
  2. A com.github.benmanes.caffeine.cache.CaffeineSpec bean is defined
  3. A com.github.benmanes.caffeine.cache.Caffeine bean is defined

For instance, the following configuration creates cache1 and cache2 caches with a maximum size of 500 and a time to live of 10 minutes

spring.cache.cache-names=cache1,cache2
spring.cache.caffeine.spec=maximumSize=500,expireAfterAccess=600s

If a com.github.benmanes.caffeine.cache.CacheLoader bean is defined, it is automatically associated to the CaffeineCacheManager. Since the CacheLoader is going to be associated with all caches managed by the cache manager, it must be defined as CacheLoader<Object, Object>. The auto-configuration ignores any other generic type.

31.1.9 Simple

If none of the other providers can be found, a simple implementation using a ConcurrentHashMap as the cache store is configured. This is the default if no caching library is present in your application. By default, caches are created as needed, but you can restrict the list of available caches by setting the cache-names property. For instance, if you want only cache1 and cache2 caches, set the cache-names property as follows:

spring.cache.cache-names=cache1,cache2

If you do so and your application uses a cache not listed, then it fails at runtime when the cache is needed, but not on startup. This is similar to the way the "real" cache providers behave if you use an undeclared cache.

31.1.10 None

When @EnableCaching is present in your configuration, a suitable cache configuration is expected as well. If you need to disable caching altogether in certain environments, force the cache type to none to use a no-op implementation, as shown in the following example:

spring.cache.type=none