Our schemas are dynamic and the application maintains the schema (this is a core feature for us and not as a workaround for "schemaless") JSON indexing is quit

Our schemas are dynamic and the application maintains the schema (this is a core feature for us and not as a workaround for "schemaless") JSON indexing is quite advanced in mongo. Postgres is catching up (gin indexes on jsonb arrays). So are aggregations on json MongoDB Atlas - fully managed and by MongoDB core committers We don't have a lot of multi document ACID requirements. We are on GCP and cloud sql is a very basic postgres deployment and we don't want to mange infra