I recently started to model an application that makes extensive use of web service calls, and my customer asked me to include some sequence diagrams to show the flow of calls in the system. This prompted me to wonder, "What is the correct way to show distributed systems in a UML sequence diagram?"

If this link is correct, you show remote systems as actors.

This seems like a decent approach, leaving me with only a slight problem in that the remote system is treated as one amorphous blob, with no way I can see to declare classes/objects within that blob. If a web service can be invoked by calls to different commands/controllers/classes, then I need a little more detail here. Time to dig into bowels of the UML specs ...