I have seen Uber ping-ponging between Postgres and MySQL. My approach to this problem is, can we use Postgres? If the answer is yes (90% of the cases) we use Postgres. If the answer if no investigate what is good out there. For k:v I have used Riak with much satisfaction. For distributed databases we are experimenting with TiDB. It looks very promising so far.

There is only one problem with Postgres that I run into. Client connection scalability. This is why many users use a connection proxy in the front of it.