Which design will give a better performance, Distributed Architecture or Clusterd? We currently run 140 - 150 active cases, daily searches and processing. We have 1 master server and 2 additional nodes.
It's a tough question to answer!
Both setups will provide improvements around review since they can utilize utility nodes for retrievals & imaging. Each node is an independent case home so they don't utilize each others resources, e.g, Node1 can't use Node2 for review on its case or offload processing to it.
In a distributed architecture, on top of having utility nodes it's possible to have processing & review roles configured across nodes meaning you can utilize all of your resources across the DA unlike in a cluster. As an example, Node1 could be using Node2 to offload processing too and be used for review to work as a loadbalancer.
I know that's not really a direct answer but I hope it gives you some guidance to make a decision. If you contact me by private message with some details, I can try put you in touch with consultancy here to help if you think that'd be useful.