To improve query job performance, what kind cluster setup is better?

Versions (relevant - OpenSearch/Dashboard/Server OS/Browser):
Opnesearch V2.11.0

Describe the issue:

Crawlering data size: 200TB; Files number: more than 20millions

Cluster node number: 4
OS: Windows Server 2022
Memory: 128GB

I’d like to know to get the best query performance,
which way is better in the below two plans?

PLAN-I:
Node-1 : Cluster Manager node
Node-2-3: data, ingest node

Plan-II:
No specified roles defined for each node;
that each nodes role is set as default: Cluster manager eligible, …

What I want to clear is that creating a Cluster_Manager node
where no index data will be created, will help improve operation performance, faster query response .
If there is other cluster configuration will improve query
operation also teach me please.
Thanks in advanced.

Configuration:

Relevant Logs or Screenshots: