Thanks for your response. This helps but I still have the following question:
If each server instance starts out "owning" all the timers does that mean there is some priority algorithm to determine which of the instances actually triggers each timer? So on average a recurring timer would be triggered evenly across all instances in the cluster, right? There is no single instance where the specific timer is bound. I'm very interested in how and if the timer executions are load balanced across the cluster.
I'm wondering if the load balancing behavior is similar to that you might find in the Quartz Scheduler, described here:
http://wiki.opensymphony.com/display/QRTZ1/ConfigJDBCJobStoreClustering
[Message sent by forum member 'tmjkeeney' (tmjkeeney)]
http://forums.java.net/jive/thread.jspa?messageID=321470