Separate index and search workloads without remote-backed storage

Versions: 3.4

Describe the issue: The documentation states that separate indexing and search tasks are possible in a cluster with remote storage enabled. Remote-backed storage use for segment replication. But segment replication also support node-to-node communication. Is it possible to allocate search nodes in a cluster without using remote storage?

Configuration:

Relevant Logs or Screenshots:

@vit According to OpenSearch documentation, the search node role is designed for remote-store-enabled clusters with segment-replication-enabled indexes.

Yes, I read that section of the documentation. But is this a requirement? As far as I understand, remote storage is only needed for replication. Perhaps a second configuration option is possible, enabling node-to-node replication?