Par Hichem Chaalal, soutenue à USTO (Oran, Algérie) le 21 janvier 2021. Co-encadrée avec Hafida Belbachir.
Classical databases represent the traditional RDBMS’s and the most widely used RDBMS in the world of databases and information systems; they have been regarded as the best systems for managing data. Today with the growth of the applications and data consumers, and its openness to the general public, traditional Databases are not able to meet the needs of a large number of applications, including OLAP data processing and Business Intelligence analysis; As a result, many variants of DBMS have emerged like: Column Store, In Memory and NOSQL Databases, that better meet users’ expectations, which are better adapted to current needs. As a result, the scope of classical databases has become increasingly restricted to handle OLTP models and other few models.
To deal with this problem, vertical fragmentation is the best way to effectively handle the OLAP model, but this technique fails to handle some analytical queries with low selectivity, presenting poor results in some cases. In this perspective, we propose a new vertical fragmentation design T-Plotter which makes it possible to deal effectively with the whole of analytical queries and improve the performance of RDBMSs to process the OLAP data models.