Optimize network calls for multiple primary keys in a set

Earlier my set had only 1 primary key for all the bins, So i was able to query like this: select * from ns.set where PK=‘primary_key’; And probably the query was from a single node which means single network call.

Now i have 2 primary keys in the set. So the query can go to 2 node clusters. That means 2 network calls. Is there any way by which i can optimise it to make 1 network call?

This will always translate to a single get that will go to the node owning the master partition (default policy) for that specific record. No matter how many nodes you have, the client will always go directly to the right node and will get the record in a single network call (other than edge cases, during migrations for example, or during retries due to network disruptions, etc…).