Is the out-of-box end-to-end visibility achieved only if you have a overall business process model designed, say to include e.g. order management, invoicing and shipping process ?.
Quite often “order processing”, “invoicing” and “shipping” business processes are kind of disconnected as we do not model them in a single process. The Order Management process model receives the PO, validates, maps it and sends it to the backend. This process stops right here. The Shipping process model kicks off on its own when the back end sends the shipping informarion to middleware and we map it to EDI and send it to customer.
Is writting custom code to link these standalone processes - the only way to get end-to-end visibility ?
The short answer is: yes, you only get end-to-end visibility if you model the overall process. Currently there is no way to link serial standalone processes into a single view that can track the status of all of them.
Instead, if you want end-to-end visibility, you would have to model this using referenced processes. You can model “order processing”, “invoicing” and “shipping” each as standalone processes that can be run independently. Then, you could define a single parent process, and drag-and-drop the “child” processes into the overall parent . At runtime, in MWS you can then see the status of the overall set of processes, and drill into them to see the status of the children as well.
Note that in our planned webMethods 8 release, we will also provide a facility to make the referenced processes dynamic – i.e. you may not know which child process to execute until runtime (based on some data or condition). This may get closer to the functionality you desire, if you don’t want to model all the process sequence up front.
You can also look at the option of correlation process which runs independently and meets back the parent process when it completes. The parent process would not have the idea of the correlation process and would only meet when it completes and publishes the correlation document.