Aside from our particulars, I just wanted to confirm if anyone ever experienced such "matched flow" behavior in the broker? As I understand, there is nothing in QPID to enforce the publisher to be as slow as the subscribers. Of course, if the publisher is slow for some other reason, the (faster) subscriptions can do no better than match its speed. However, this doesn't explain why publisher speed drops down "naturally" in the "slow" brokers if we run the test with higher number of subscriptions.
Thanks, Andy
