What is the effect of load-blancing, and clustering on the number of Jdbc connections ?
Jdbc pools are created per partition (ie JVM) process. If you have multiple partition processes running, it is important to note that each partition will open a separate set of connections. So, the total number of Jdbc connections would be the summation of connections in each of the pools in different partitions. In general, it is best to deploy your tightly coupled EJBs as part of a single J2EE application. This will guarantee good performance, better resource utilization, and also can avoid a two-phase commit protocol where it may not be needed . Please note that, the Jdbc connections are opened only when they are needed; hence if an application is replicated for fail-over, the connections would be created when that replica is actually used.
