I just did DV2 training last week with Dan Linstedt and Cindy Meyersohn, and it was great! I understand this methodology a lot better now. I will be taking my certification exam very soon.
I found a case (when trying to build a view on this data) where I think we broke the "unit of work" with our RDV model for a file we are importing that contains some counts of events we want to capture. We definitely missed a hub in the initial design, but we also got the grain of the data wrong across files. I came up with a proposed new RDV model, as did our current architect. We've been discussing when a satellite should hang off a link, vs when you should create an additional hub.
I would love your feedback (from the group at large, and of course Dan in particular) as to how you make this design decision. So in the attachment, the "traditional" proposed design is mine, the "ensemble" one is the architect's.
My thoughts were that you would want the satellite to hang off the link if that is the right granularity of the data, and that the additional hub at that grain would make the RDV structures unnecessarily heavy. There is a possibility (but not a certainty) that there is other data that would be at this same grain and would thus also become an additional satellite off of this link. Does there come a point when, if you have several satellites with a parent link, that it should have been a hub instead?
PS: I told the architect I was posting here, so this is not a "but Dan Linstedt says this" scenario. He's very interested to hear the recommendation and why. Thanks in advance for your feedback!