You need to prevent Dynamic Optimization from placing any of the three virtual machines in the same node
You need to ensure that App1 fails over automatically to another server if a single node in Cluster1 fails