In modelling of our Systems Architecture we have elements representing the physical parts of our system. These physical parts are connected via interfaces – one interface can connect 2 or more physical parts. We also have a representation for the logical parts of the system. These logical parts are allocated to physical parts. These logical parts can also be seen to have interfaces.
What troubles me, is that we seem to be defining the interfaces between logical functions as being the same as interfaces between physical parts. Although, you can say that the logical does use the physical, I’m not so sure that they are the same thing.
The other approach is to define the ‘logical’ interfaces as a design artefact and then link them to the physical interface. I’m also not convinced this is the correct approach. As an example, although tcp/ip traffic passes through a copper cable, it is only the transmission mechanism. The decoding of that data is in the controller (and more correctly, in the software stack).
I’m wondering how others have approached this type of scenario?