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 | |
---|---|
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 | |
---|---|
You can also use the standard JSR-107 (JCache) annotations (such as
|
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.
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 | |
---|---|
It is also possible to force a particular cache provider by setting the
|
Tip | |
---|---|
Use the |
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 | |
---|---|
In the preceding example, an auto-configured |
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.
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 | |
---|---|
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 | |
---|---|
Spring Boot has general support for Hazelcast. If a
single |
There are two ways to customize the underlying javax.cache.cacheManager
:
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 | |
---|---|
If a standard |
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
Spring Boot has general support for Hazelcast. If a
HazelcastInstance
has been auto-configured, it is automatically wrapped in a
CacheManager
.
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 | |
---|---|
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. |
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.
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 | |
---|---|
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 |
Tip | |
---|---|
You can take full control of the configuration by adding a |
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):
spring.cache.caffeine.spec
com.github.benmanes.caffeine.cache.CaffeineSpec
bean is definedcom.github.benmanes.caffeine.cache.Caffeine
bean is definedFor 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.
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.
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