My thoughts on the case for a SINGLE database. My research started as a entire-surname study, and then evolved to include ancestors and descendants of cousins.
I've now been researching for over 28 years, and my database includes over 395,000 individuals - approximately 83%of them connected in a single main tree. The remaining 17% are from side projects to study friends' genealogy, to explore certain historical families, to explore certain historical themes (innovators and scientists), etc. As I've said many times, in order to believe my research conclusions, one must be able to trust my sources, so I make an effort to document every fact/event with a source & citation. I'm a source lumper, as opposed to an extreme splitter, and I currently have over 13,000 sources and over 2,200,000 citations, from 1,800 repositories. Yes, I spend time nearly every day, usually several hours per day, on this hobby.
By storing everything in a single database, I have been able to show how my various relatives are related (ninth-cousins who are married, for example) as well as showing connections in various ways to Europe's ruling dynasties. When I share reports with the younger generations, they really enjoy seeing their historical connections. If I had kept separate databases for each of my side projects, those connections might not have been documented.