Run model evaluation in multiple servers

Hello i am using UQlink to run a model using an external software (in a docker container). Indeed, i was reading the documentation as regards the dispatching module, yet I am at quite a lost about how to solve my specific problem. I have 3 servers in a local network, and I am running an inversion study, the goal is to evaluate the model, using the 3 servers (async), as to maximize the running time in the three servers. Thanks for the help,
Andres Elias