I don't know if LevelDB is better in regards to startup time - I believe it is. If there is rarely any backlog of messages, brokers come up within seconds.
If you really want sub-second message passing performance through a SPOF (single-point-of-failure), you might consider an alternative solution. Such as setting up two independent brokers and having all messages duplicated on both. That would require de-dupping on the consumer end, but that's always a possible scenario with JMS anyway. Honestly, once the brokers are up and functional and the solution is well built, the brokers will run for months (or longer) without problems. Then the only significant issue is frequency and impact of scheduled maintenance. Of course, the business need for timeliness is really the driver here. Without knowing the business, it's hard to know how if these generalities will work for you. If the business needs sub-second message delivery 90% of the time, and can handle delays of up to a few minutes, you should be able to achieve that. If sub-second message delivery is needed 99.999% of the time, it will be a challenge. -- View this message in context: http://activemq.2283324.n4.nabble.com/Failover-and-non-persistent-messages-tp4676862p4676975.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
