Denormalization is specially focused on enhancing performance when doing inquiries with a way to attempt a performance elevation of a system through data duplication in DB design. However, more time is needed for data entry, a revised city, not to mention what gives up the data consistency that is a merit of the DB, and this way is not helpful to performance elevations of total systems with data duplications. In this thesis, we design each model of normalization and denormalization to indicate a problem in denormalization, which most domestic companies apply with a purpose that increases answer time and measures it as a total system along with database size, the number of table joins, the ratio of reading and writing, and inquiry frequency under the same conditions. Data duplication through denormalization causes approximately an 8% increase in wasted storage while system answer speed drops by approximately 30% on average due to side effects of duplication (if based on a paper). These side effects increase when the database size increases, the ratio of reading and writing and system users also increase markedly. The data duplication permission method is negative in a general information system so renewal coincides to an inquiry in total system performance. For these reason, this thesis emphasizes the needlessness of denormalization for general company information systems as sales systems.