database sharding strategy, which approach works best

hey everyone, i've been thinking a lot about scaling databases lately, specifically sharding. we're anticipating significant user growth and need a solid plan for horizontal scaling. i've looked into a few strategies: hash-based, range-based, and directory-based. each seems to have its own pros and cons regarding query performance, rebalancing, and operational complexity. hash-based can lead to uneven distribution if not done carefully, range-based can cause hot spots with time-series data, and directory-based adds another layer of indirection. what have you all found to be most effective in production? are there other sharding strategies i should consider, or common pitfalls to avoid when implementing one?