This was clickbait for me because last year I implemented an ISAPI DLL for a web server doing lookups against a SQLite database. But I was disappointed the art

This was clickbait for me because last year I implemented an ISAPI DLL for a web server doing lookups against a SQLite database. But I was disappointed the article seemed to immediately decide to implement a solution for a problem that is barely described. He's got a growing database that SQLite can handle, but "the server I am renting will not". What kinds of transactions are coming in/going out? How many transactions per second? How big are they? What response time is required? Since none of that is specified, worrying about sharding smells a bit like "the cart before the horse". Indeed, his final concern about "over-engineering" seems appropriate, and I think the author's fascination with implementing sharding is the solution for which he might not yet have the problem.