The larger the number of tables, the more additional L/O operations and processing logic required to join them Therefore, tables are sometimes denormalized to yield less L/O in increase processing speed Unfortunately, with larger tables, you pay for the increased processing speed by making the data updates less efficient, by making indexing more cumbersome, and by introducing data redundancies that are likely to yield data anomalies In the design of production databases, use denormalization sparingly and cautiously.