For reference, here is Gerben’s attempt at modelling VM:
…and his narrative:
The blue Composition relations and the route via the VM Image Artifact are equivalent to what has been used in View 194 [In this view…it is also not a bad choice to use Composition as the relation between the System Softwares of a virtual machine, as it mimics the structure of the Artifacts that Realize the System Software elements.].
The other alternatives shown are:
• The orange Assignment shows the relation between the active parallelization host element and the virtual device.
• The green Serving shows that the virtual device requires the parallelization host to function.
• The red Realization is derived from the route via the VM Image Artifact.
There are of course many, many more ways to model hosted virtual servers. There is a myriad of choices in what you show, leave out, etc. Here again, it becomes clear that you need to
find your own pattern based on what you like and need. A small reminder: this book is in the end about understanding the language and which options it gives you.
This seems consistent with Peter’s model where multiple virtual nodes compose the VM Host. However, as Steve points out, there needs to be a way to aggregate/compose system software and devices into a node, therefore the dependency model breaks when analysing dependencies on the consitituent components of a node.
