Apache ActiveMQ issue

Our setup is as follows: 1) We have our cloud servers on Rackspace (Moving to AWS this month) 2) There is a Client local installations Software installed at 250+ locations. 3) The Client local installations will send data to the Cloud on a real time basis. 4) The Cloud Servers will also send data to POS ( configuration etc). There is a Duplex Queue between the Client local installation and the Cloud Server. Also, there is a Virtual Topic setup on the Cloud server which we use to send data down to the Client local installations. We are facing some issues in ActiveMQ : 1) The KahaDB @ our company cloud servers gets corrupted sometimes. (Infrequently) 2) KahaDB @ Client local installations get corrupted frequently ( Mostly due to power fluctuations) 3) The Broker at the Client local installation end ( and sometimes at the server End) freezes - it does not send/ receive any data. We need to restart it to start the communication again. 4) Sometimes, the Broker at the Client local installation end throws an error in the logs saying Destination Host unreachable even if the Client local installation is connected to the internet and the Cloud Server IP can be pinged. We need to restart the Client local installation MQ or the Cloud MQ to get the communication started.