Databases are really just data structures and algorithms along with some notion of durability. Client/Server databases are just remote data structures. (E.g. R

Databases are really just data structures and algorithms along with some notion of durability. Client/Server databases are just remote data structures. (E.g. Redis is short for "Remote Dictionary Server") Sometimes you want your data structures and algorithms to run locally. Could be performance, privacy, cost, or any number of reasons. Local, in-memory data structures hit a few bottlenecks. First, they may not fit in memory. A mechanism for keeping the dataset in larger storage (e.g. disk) and paging in the necessary bits as needed extends the range of datasets one can comfortably work with locally by quite a bit. That's standard SQLite. A second potential bottleneck to local data structures is distribution. We carry computers in our pockets, on our watches, in our cars. Delivering large datasets to each of those locations may be impractical. Cloud based VFS allows the benefits of local data structures on the subset they need without requiring them to fetch the entire dataset. That can be a huge win if there's a specific subset they need. It always depends on the use case, but when the case fits there are a lot of big wins here.