I’m not even sure what we are debating. Spark is for big data work, it isn’t something you would typically use as a general purpose database backing your application. My original point way back is that expecting customers to learn an entirely new paradigm for storing and querying data is a poor decision, and limits your work to niche use cases. Spark has a large niche but it isn’t comparable to MySQL / Postgres or other newish data stores like Dynamo.

Furthermore any competent engineer knows SQL because ORMs are cumbersome and annoying for anything except basic use.