Is data integrity strictly a microservices problem? I swear I've seen a ton of monoliths with not-demoralized databases where it's fairly confusing as to what column is correct since there's a bunch of conflicting entries.

I'm a bit surprised that data migration is considered a problem for microservices. Is any "class" in the application allowed to create SQL statement? It seems like it'd be super easy to lift db functions to a microservice and replace the monolith's SQL calls with HTTP (?) calls.