java.util.concurrent package provides data structures such as AtomicLong, CountDownLatch, ConcurrentHashMap, etc. which are useful when you have more than one thread reading/writing data to the data structure. But to provide thread safety, all of these threads are expected to be on a single JVM/machine.

There are two major benefits of distributing data structure −

• Better Performance − If more than one machine has access to the data, all of them can work in parallel and complete the work in a lesser timespan.

• Data Backup − If a JVM/machine goes down, we have another JVMs/machines holding the data

Hazelcast provides a way to distribute your data structure across JVMs/machines.