How do you think about schema design?

Schema design balances normalization, performance, and future evolution. I start by identifying entities and relationships, then apply normalization rules to eliminate redundancy while ensuring data integrity. For analytical workloads, I denormalize into star or snowflake schemas, creating fact tables with foreign keys to dimension tables. I choose appropriate data types, enforce constraints, and add indexes on foreign keys and frequently queried columns. Partitioning large tables by date or geography improves query speed and maintenance. I also plan for schema evolution: using nullable columns, versioned tables, or change‑data‑capture mechanisms to avoid breaking downstream consumers. Documentation and naming conventions are critical for maintainability. Throughout, I review the expected query patterns and adjust the design to optimize read/write balance.