Here is a recording of an old presentation that discusses both the 1.x and 2.x architectures with detailed diagrams starting around 25min into the video.
https://www.youtube.com/watch?v=bZPpt4NnvsA The slides on threading/execution model are more future looking and not yet in 2.0. The final perf numbers for 2.0 improved further. Some of this info was captured in the first design doc listed under https://issues.apache.org/jira/browse/STORM-2306 TOPOLOGY.DISRUPTOR.BATCH.SIZE is the number of items that are batched together and inserted in one go into any disruptor queue .. unless a flush timeout occurs. In 2.0 it is replaced with `topology.producer.batch.size` and `topology.transfer.batch.size` (for remote transfer) and described here: https://storm.apache.org/releases/current/Performance.html -roshan
