Which schema design—star or snowflake—provides better query performance, and how does storage impact this choice?

Star schemas are denormalized, so a query that aggregates fact data with dimension attributes typically requires only one or two joins. This reduces CPU time, I/O, and query planning overhead, making star schemas faster for most OLAP workloads. Snowflake schemas normalize dimensions into sub‑tables, which means a query must perform additional joins to assemble the full dimension. Those extra joins increase CPU cycles and can cause more disk seeks, especially if the dimension tables are large. However, snowflake schemas use less storage because they eliminate duplicate dimension rows. If storage cost is a major concern or the dimension data is very large and rarely queried, a snowflake may be preferable. In most BI scenarios, the performance benefit of a star schema outweighs the storage penalty, so star schemas are the default choice for reporting.