Replicating is the process of having copies of the data across different locations. This enables us to have a bigger throughput as we can have different services serving the same data. It also brings complexity, at least for the database servers, as they need to keep everything synchronized:

Doing replicas also gives you redundancy in case of any failures in regards to parts of the server nodes, depending on your configuration. If you use a database cluster for more than one application, you don't need to have replicas of a database in all cluster nodes; it all depends on your needs.