In {productname-short}, you can run a distributed workload from a notebook or from a pipeline.
You can run distributed workloads in a disconnected environment if you can access all of the required software from that environment. For example, you must be able to access a Ray cluster image, and the data sets and Python dependencies used by the workload, from the disconnected environment.