Scalability can be subjective for every deployment. You may be looking for higher query performance, higher availability, or higher indexing performance. In some cases, for small deployments, it may also be sufficient to just add capacity to the existing node in order to scale to higher performance or more volume. However, there is a limit to how much you can scale using a single machine.
A cluster is a group of servers that functions as a single system in order to leverage parallel processing and make the system highly available for its clients. You can simply start multiple Elasticsearch servers with the same cluster name in a dedicated bare metal box or a virtualized environment. These instances will form a cluster. Elasticsearch is designed to provide linear scalability so that in order to achieve more scalability, you can simply add more nodes to the cluster. Elasticsearch has built-in discovery and replication...