Yeah what i was thinking. There is no state to be managed. Its just a pipeline which doesnt not anything more than more machines to handles more load. Its relatively easy to implement.