Elasticsearch allows you to setup heterogeneous clusters, that is, nodes with different configurations within the same cluster. Elastic (the company) refers to this architecture as “Hot-Warm”, but it’s called tiered storage if you come from a storage background.
The canonical example is that you have a bunch of data you want to keep online and able to query, but it becomes less relevant over time. You want to cut costs, so you have your “Hot” data that is written and/or read frequently, most likely on SSD, and then “Warm” data accessed less frequently on less expensive nodes, most likely on spinning disks. But it doesn’t end there, this architecture can be used and extended in different ways