Postgres is our goto, but in particular we have been pretty satisfied lately with AWS so-called “serverless” Aurora Postgres, which allows utilization to scale down during quiet periods, which is very helpful in a lots-of-microservices context, which can otherwise have a pretty high cost floor from having many Postgres DBs that have to be sized to accommodate their peak load.

DynamoDB, which I was slow to come around on, is attractive for the same reason, although it only works if the use case fits, obviously.