[AYBABTME]

I'm curious what your environment is like, because my experience has been the opposite. No one that I know would recommend a nosql database in the vast majority of cases. Especially since there's great offerings out there nowadays, and plenty of proven experience with stuff like MySQL (most often with Vitess) supporting super high loads such as Youtube, Shopify, GitHub, etc. MySQL scales for a lonnnnng time. And the same is true for Postgres, with multiple hosted offerings and also reimplementations that are PG compatible.

My experience is that NoSQL databases are used in last resort situations: you _need_ Cassandra, etcd, ElasticSearch for XYZ reason, as a dependency to another distributed system. But I haven't seen anyone suggest actually using stuff like MongoDB and the likes.

[audioheavy]

There's definitely interest in both, but oftentimes for what I would argue are the wrong reasons. MySQL and PostgreSQL scale to a point but with complexity and data isolation limitations, but they will be always popular because of their longevity. Some NoSQL options scale better horizontally but the limitations of isolation levels are even more significant. Just because most of them now can handle ACID transactions and some form of joins does not mean they can handle strict serialization at scale. There are some, like Fauna, that do, however.

[AYBABTME]

My experience is that MySQL scales as high as you need it when coupled with Vitess, because most data schemas can be sharded, and you can scale each shard vertically enough for all the workloads I've seen.