That's why I choose Postgres: aside from extreme scale or latency, you can throw pretty much anything at it. Also, given expert advice, its performance degrades gracefully i.e. lots of easy ways to optimize and workaround issues.
Example: pg's type system is insanely flexible incl native support for high speed JSON, latlng, geometry, and more - and types are "batteries included" incl matching, inspection/extraction (e.g. JSON navigation), conversion, concurrency and concurrency control, ACID recovery, replication, etc. For example, it's one way line of code to create an index on the results of a JSON navigation expression whose result is then converted to another datatype e.g. timestamptz. Performance is then 100x+ on queries that use that expression.
Pg datetime handling is also world class incl timezones and conversion, query and manipulation functions. Want the records from last Tuesday across timezones, given data in seconds-since-the-epoch? The SQL is shorter than in English.