Replicated Cache with Embedded Infinispan

What You Will Learn

How to create a clustered Infinispan cache in replicated synchronous mode, where every entry is copied to all nodes in the cluster.

Prerequisites

Java 17+

Step 1: Add the Embedded Dependency

Add Infinispan core to your pom.xml:

<dependency>
   <groupId>org.infinispan</groupId>
   <artifactId>infinispan-core</artifactId>
</dependency>

Step 2: Create a Replicated Cache

Set up a clustered cache manager and configure a cache with REPL_SYNC mode. Every entry is synchronously replicated to all nodes:

      // Setup up a clustered cache manager
      GlobalConfigurationBuilder global = GlobalConfigurationBuilder.defaultClusteredBuilder();
      // Initialize the cache manager
      cacheManager = new DefaultCacheManager(global.build());
      // Create a replicated synchronous configuration
      ConfigurationBuilder builder = new ConfigurationBuilder();
      builder.clustering().cacheMode(CacheMode.REPL_SYNC);
      Configuration cacheConfig = builder.build();
      // Create a cache
      cache = cacheManager.administration()
              .withFlags(CacheContainerAdmin.AdminFlag.VOLATILE)
              .getOrCreateCache("cache", cacheConfig);

Step 3: Populate and Query the Cache

Store entries and observe that each node holds all the data. Use SKIP_REMOTE_LOOKUP to see the local view:

      // Store the current node address in some random keys
      for(int i=0; i < 10; i++) {
         cache.put(UUID.randomUUID().toString(), cacheManager.getNodeAddress());
      }

      // Display the current cache contents for the whole cluster
      cache.entrySet().forEach(entry -> System.out.printf("%s = %s\n", entry.getKey(), entry.getValue()));

      // Display the current cache contents for this node
      cache.getAdvancedCache().withFlags(Flag.SKIP_REMOTE_LOOKUP)
              .entrySet().forEach(entry -> System.out.printf("%s = %s\n", entry.getKey(), entry.getValue()));

In replicated mode, both views show the same entries because every node has a full copy of the data.

Step 4: Run the Tutorial

mvn package exec:java

The local and cluster-wide views will display the same entries, unlike distributed mode where data is partitioned.

What’s Next

Compare with distributed mode where data is partitioned
Try invalidation mode for locally-managed data
Use distributed streams for parallel processing