Deploying Specific Services on Targeted ray worker nodes or clusters

I have developed a multi-model serving application that utilizes the @serve.deployment functionality to deploy long-running services. However, I am encountering a challenge in deploying specific services on targeted worker nodes or clusters within the Ray framework.

Currently, I am seeking a solution that allows me to deploy a particular service on a designated ray worker node or cluster without kubernetes or etc. In certain cases, I may also require deploying multiple replicas of the same service on different worker nodes or clusters by leveraging the num_replicas parameter. However, I am unsure of the appropriate steps or configuration settings to achieve this level of deployment control.

I would greatly appreciate your professional guidance on how to accomplish this objective effectively within the Ray framework. Any insights, best practices, or specific configuration examples that you can provide would be of immense help in resolving this matter.