Why should I use replication? What benefits does it bring to my cluster?
Replication's advantages are in the name: it replicates your data for redundancy and durability throughout the cluster. There are two types of replication: high-availability and database replication to another cluster.
When HA is enabled (High Availability/redundancy_level=2), SingleStore distributes the data across your cluster so that each leaf node has a "backup" or redundant leaf node which contains the same data. This allows you to have a second leaf ready in case of any issue with the original leaf node.
When enabling HA, we recommend putting the two nodes that hold the same data on different hosts. This gives you the advantage of always having a quick and automatic "failover" to another node on a separate machine should the machine you have your "primary" node (the machine with the version of the node that is accessed by default) experiences a hardware failure, leaf failure, or any issue that keeps you from accessing the data on that node. In this event, SingleStore will switch over to the "secondary" node wherein all the data has been replicated to. This allows your cluster to continue functioning as you fix or replace the machine with the original "primary" node on it.
Database Replication is similar to this in that you can set up an entire secondary cluster to replicate copies of the database to a backup disaster recovery cluster. In this instance, the second cluster is hosted on a different set of hosts altogether, so if the first cluster needs to be brought down entirely, you can switch your endpoints over to the second cluster.
You can read more about high availability here, how to enable and manage it within our docs here, and replication to another disaster recovery cluster here.