Database Sharding versus Read Replicas, you know, when should you pick one over the other?

hey everyone, ive been thinkin a lot about scaling databases recently. were hittin some pretty heavy read loads on our main app database, and im weighin the options. sharding seems like the ultimate solution for distributin both read and write traffic, but it comes with a ton of complexity in terms of app logic and management. read replicas, on the other hand, are much simpler to set up and handle read-heavy scenarios well. but they dont help with write bottlenecks and can introduce replication lag. so, im curious: at what point do you all find yourselves leanin towards sharding over read replicas, or vice-versa? whats the key stuff that pushes you to one side of that trade-off