I have an application which uses a launchClient java client to obtain a list of jobs from a database via the application server. The client then divides the list of jobs, a list of numeric ids, into chunks and then invokes stateless session beans deployed to the application server over RMI. This works very well with network deployment workload management (WLM) in terms of distributing the workload. I wanted to simplify the client as much as possible, by not having to retieve the initial list of jobs from outside the app server. However:-

1. Am I correct in thinking that WLM only works for client requests that originate from outside of the network deployment cell ?.

2. That if I am correct on my first point, people get round this by partitioning the workload up and sticking it on a queue via JMS, which the nodes in the cell collect.