If https://hazelcast.com/[Hazelcast] is on the classpath and a suitable configuration is found, Spring Boot auto-configures a `HazelcastInstance` that you can inject in your application.
Spring Boot first attempts to create a client by checking the following configuration options:
* The presence of a `com.hazelcast.client.config.ClientConfig` bean.
* A configuration file defined by the configprop:spring.hazelcast.config[] property.
* The presence of the `hazelcast.client.config` system property.
* A `hazelcast-client.xml` in the working directory or at the root of the classpath.
* A `hazelcast-client.yaml` in the working directory or at the root of the classpath.
NOTE: Spring Boot supports both Hazelcast 4 and Hazelcast 3.
If you downgrade to Hazelcast 3, `hazelcast-client` should be added to the classpath to configure a client.
If a client can't be created, Spring Boot attempts to configure an embedded server.
If you define a `com.hazelcast.config.Config` bean, Spring Boot uses that.
If your configuration defines an instance name, Spring Boot tries to locate an existing instance rather than creating a new one.
You could also specify the Hazelcast configuration file to use through configuration, as shown in the following example:
Otherwise, Spring Boot tries to find the Hazelcast configuration from the default locations: `hazelcast.xml` in the working directory or at the root of the classpath, or a `.yaml` counterpart in the same locations.
We also check if the `hazelcast.config` system property is set.
See the https://docs.hazelcast.org/docs/latest/manual/html-single/[Hazelcast documentation] for more details.
NOTE: Spring Boot also has <<features#features.caching.provider.hazelcast,explicit caching support for Hazelcast>>.
If caching is enabled, the `HazelcastInstance` is automatically wrapped in a `CacheManager` implementation.
