I am curious to know (aside from obvious theoretical benefit) of Wikibase FEDERATION - What are known (good-bad) PRACTICES: if they and how they came into reality?
Wikibase federation has been under development for some time. It’s not yet complete and Wikibase Cloud dropped support for it in the meantime.
In my opinion, it is a massive engineering feat and something worth pursuing by a dedicated team. At this point you are not simply creating a graph database, but a graph database network, and building a network carries with it many technical challenges.
In the meantime, my idea to approximate this experience is having multiple graphs in the same query service, with SPARQL queries written to join the datasets together as needed. This is how I am planning on incorporating Wikidata’s data in Librarybase queries.