1. On two machines (called "paddington" and "yogi" in my tests) install A-MQ ER12. 2. Run "bin/artemis create temp" on each 3. On the machine that is to be master ("paddington" in this case) replace the etc/broker.xml file with paddington.xml below. 4. On the machine that is to be slave ("yogi" in this case) replace the etc/broker.xml file with yogi.xml 5. Edit the two broker.xml files to have meaningful hostnames and/or port numbers, if necessary 6. Start the two brokers 7. Use any suitable JMS client to send a single message to any destination on the master broker, whilst monitoring the network traffic to the slave. The client should be capable of timing how long the send operation takes (I can supply a client if necessary). The contents of the message should appear on the wire between the master and slave – this is a way to be sure that the message has actually been forwarded 8. Once messages are being forwarded correctly, add one second of delay to all network traffic between the master and the slave. It doesn't matter for this test if the delay is added to everything. tc qdisc change dev eth0 root netem delay 1000ms 9. Check the delay is really working: kevin@paddington source]$ ping yogi PING yogi (192.168.1.70) 56(84) bytes of data. 64 bytes from yogi (192.168.1.70): icmp_seq=1 ttl=64 time=1116 ms ... 10. Retry the JMS message operation. If it takes less than two seconds (one second for a network packet in each direction), then the send on the master has returned to the client before the slave has received and confirmed the update.