We are using JBPM in an application that is aggregating data from a variety of legacy systems. User tasks will interact with data from and write data to these legacy systems. Given this environment, what role should JBPM task variables play? It seems to me that I will not want to create process variables that gets stored in the JBPM database as well as in the legacy system.
What component is advisable to use to retrieve and store data? At what point in the process definition (TaskControllerHandler?)? How would the business analysts, then configure the tasks with the data fields required?