I'd say that A SQL variation is probably the right first choice for most software projects where server stored data is involved. As much as I enjoy other options (Mongo, Cassandra, BigTable, Dynamo, etc) in different scenarios.
A lot of this will come down to many developers do use SQL first... Especially in Java and C# circles (corporate it developers in particular). So staying closer to what people are familiar with has value.
I will generally push for PostgreSQL or CockroachDB over other SQL varieties though. MS-SQL is okay, Oracle can be a pain, both being costly. Not a fan of Maria/MySQL only because of old behaviors, and every time I've used it, there's something annoying (utf8 isn't, as an example).
In either case, a SQL database service can often scale to the low millions of users if you're pretty good with how you structure things... A poorly constructed database and application can generally scale at least to thousands of users without issue. As most application development are internal business applications, that's usually enough.
A lot of this will come down to many developers do use SQL first... Especially in Java and C# circles (corporate it developers in particular). So staying closer to what people are familiar with has value.
I will generally push for PostgreSQL or CockroachDB over other SQL varieties though. MS-SQL is okay, Oracle can be a pain, both being costly. Not a fan of Maria/MySQL only because of old behaviors, and every time I've used it, there's something annoying (utf8 isn't, as an example).
In either case, a SQL database service can often scale to the low millions of users if you're pretty good with how you structure things... A poorly constructed database and application can generally scale at least to thousands of users without issue. As most application development are internal business applications, that's usually enough.