We should handle something like (minimum/maximum) delay in async topology for some topologies that guarantee maximum delay to receive messages.