Two Database nodes

OpenSearch
1

Hi all,

I’m setting up OpenSearch and followed the documentation, which suggests creating a coordinator node, a manager node, and two database nodes. However, I’m struggling to understand the reasoning behind having two separate database nodes.

Could someone explainwhy I have to create 2 database nodes and the benefits or the specific use case for having two database nodes in this configuration? How does it impact performance, reliability, or scalability in an OpenSearch setup?

Thanks in advance for the help!

2

How does it impact performance, reliability, or scalability in an OpenSearch setup?

It helps with reliability. OpenSearch is a distributed system and is fault tolerant when a single node fails. Indices in OpenSearch are composed of shards and by default, each index has a primary and a replica shard that are split amongst the data nodes in a cluster. If one node fails, then a shard still exists in the cluster where search requests can be routed.