Reload, Reduce and Distribute are different stages of a task. Each one requires a reload engine to open the source document, do something with it and save/distribute it to the same or a temporary document, or to the QVS. For large documents, each one may take a lot of time to finish. And may create queues when not enough engines are available at the same time.
Do as Tim said. Disable the Reduce and Distribute phases on the respective tabs, enable the reload tab and define a trigger. To get even better performance, wipe the data model at the end of the script in your QVD Generators.