How should database partitioning be applied in a production system?

Select keys from access patterns, plan redistribution, understand cross-partition queries, and monitor skew.