Have the hub (upon api invocation) send a ping to all modules, including the time sent. Each module responds with both the incoming timestamp and the time it received the ping. The data will let us estimate queue-length and load related latency factors.
Yeah, let's talk about this when it's convenient.