Assuming the light-weight version of the application was configured at a different port on the same server or on a completely different server address, you might be able to put the lightweight application services in the same pool as the normal application services, and use Priority Group Activation to activate the lightweight-version pool member(s) only when a configured connection limit was reached on a normal-version pool member. Persistent connections may still go to the normal-version server that has the connection limit, but new connections would be load balanced to the light-version.