when to use read replicas compared to sharding for databases?

hey everyone, i'm trying to figure out the best way to scale our read-heavy application. we're hitting limits with our current single database instance. i'm weighing the pros and cons of adding read replicas versus implementing database sharding. read replicas seem simpler initially for distributing read load, but i'm concerned about replication lag and eventual consistency. sharding feels more complex upfront but might offer better long-term scalability and performance. what are your experiences? when did you decide to go with one over the other, and what were the key factors in that decision? any common pitfalls to avoid