Session Replication with Spring Boot and Hazelcast

What You’ll Learn

In this tutorial, you’ll deploy two Spring Boot applications which use Hazelcast for session replication.

Before you Begin

JDK 1.8+
Apache Maven 3.2+

Spring Boot Application

Let’s have a simple web application with the following controller and the HTML template.

WebController.java

@Controller
@RequestMapping("/")
public class WebController {

    @RequestMapping(value = "/")
    public String index(HttpSession httpSession) {
        Integer hits = (Integer) httpSession.getAttribute("hits");
        if (hits == null) {
            hits = 0;
        }
        httpSession.setAttribute("hits", ++hits);

        return "index";
    }
}

index.html

<!DOCTYPE html>
<html xmlns:th="http://www.thymeleaf.org">
	<body>
		<h3>Session Id</h3>
		<p th:text="${#httpSession.id}"/>
		<h3>Hits</h3>
		<p th:text="${#httpSession.getAttribute('hits')}"/>
	</body>
</html>

When you start the application and open the browser at http://localhost:8080, you should see a piece of information with the current session ID and the number of hits (number of times you refreshed the browser).

Now, the problem is that if you start two instances of your application, then the counts for each of them would be completely separate. That is because each Spring Boot application replica stores its sessions separately. What’s more, if the application crashes, the session data stored in that instance will be lost. That is what we want to change with the use of Hazelcast.

Using Hazelcast for Sessions Replication

To use Hazelcast for session replication in your Spring Boot application, you need to:

add hazelcast-all dependency
add Hazelcast configuration
add WebFilter configuration bean

Here are the added parts in pom.xml and Application.java.

pom.xml

<dependency>
    <groupId>com.hazelcast</groupId>
    <artifactId>hazelcast-all</artifactId>
    <version>${version.hazelcast}</version>
</dependency>

Application.java

@Bean
public Config config() {
    return new Config();
}

@Bean
public WebFilter webFilter(HazelcastInstance hazelcastInstance) {

    Properties properties = new Properties();
    properties.put("instance-name", hazelcastInstance.getName());
    properties.put("sticky-session", "false");

    return new WebFilter(properties);
}

This way, we defined a Hazelcast configuration and used an embedded Hazelcast instance to enable session replication.

Starting the Application

To start two replicas of the application, run the following commands.

mvn spring-boot:run
SERVER_PORT=8081 mvn spring-boot:run

You should see in the logs that both embedded Hazelcast instances formed a cluster.

Members {size:2, ver:2} [
        Member [172.24.148.110]:5702 - e01c5b53-4dd8-492c-974c-7f4a7e195fb8 this
        Member [172.24.148.110]:5701 - 27305639-b75c-4f41-84e5-6cd87995511b
]

Test the Application

Open the browser at http://localhost:8080. You should see the following content.

Now, open the browser at http://localhost:8081. You should see that the hit has increased.

That means both applications share the same session data store.