database sharding strategies for high-volume data

Hey everyone, I've been trying to figure out how to scale our user data storage. We're getting to the point where one database instance is becoming a bottleneck for reads and writes. I'm thinking about sharding, but I'm not totally sure which strategy is best. We've looked at hash-based sharding and directory-based sharding, but they both seem to have their own issues with query complexity and rebalancing. What have your experiences been with different sharding methods? Have you hit any common problems or found any unexpected advantages?