| Abstract [eng] |
Majority existing programs were created or are still being developed using relational database management system technologies. Relational databases are usually normalized – tables and fields are organized respectively. This process ensures better data integrity, however multiple table joins are used when retrieving data. Join is a slow operation, therefore database is often denormalized. There are various denormalization methods in which data is organized in such a way that would be easier accessed to queries. Most often number of tables that must be scaned in queries is reduced, thus reducing the number of joins in query. These methods include joining tables into one, adding duplicate columns, adding duplicate fields and storing calculated fields. Also, during denormalization process, the table can be split to reduce table size. Although denormalization is often applied in practice, there is still a lack of clear strategies in this area. In existing researches the denormalization impact on the operation of the database is indicated only in a theoretical basis. The essence of this work is to test denormalization methods and to determine their impact. 4 experiments were carried out in the work, during which time and other parameters before and after the optimization were measured. Measurements are performed with three data sampling – small, medium and big. Summaries are provided in the results section. |
