There are tradeoffs with both design, I know that I use secondary indexes in quite a lot of my queries, so the way Postgres handles them are much better in my c

There are tradeoffs with both design, I know that I use secondary indexes in quite a lot of my queries, so the way Postgres handles them are much better in my case... but in someone else's use case (Uber) it could be terrible for performance.