Note that there is a BIG difference in the cost of a 'Round Trip' to the database in something like a classical server-based database and SQLite, which is a critical difference. With a server-based database, there are significant communication delays in the 'round trip', reaching out to it and getting a response. With SQLite, there is basically no trip to speak of, as SQLite sits in your process space. The whole concept of having lots of code to keep a cached version of the database can become a pessimism when you take into account that it will eat up process memory that could otherwise have been used to up the memory cache for SQLite and the OS, rather than building a duplicate cache for the data that is likely sitting in those already.