The primary advantage of clustering is service availability. Without the traditional PRIMARY-BACKUP architecture where the process of switching from one to the other takes time and can affect multiple users, the 'swarm' will respond immediately to all new requests, and since multiple 'bees' are responding to each request, you preserve data integrity in case of failure.
Cluster is to an application as SMP hw is to a thread. There's less bottleneck.
With a single threaded app so you see an improvement on an SMP rig?
Cheers as a learning experience, for sure. Practically, it may not improve performance