Les index pour les entrep\^ots de donn\'ees : comparaison entre index arbre-B et Bitmap

TL;DR

This paper compares B-tree and Bitmap indexes in data warehouses, demonstrating through experiments that Bitmap indexes are more advantageous for query performance.

Contribution

It provides a comparative analysis of B-tree and Bitmap indexes, highlighting the benefits of Bitmap indexes in data warehouse environments.

Findings

Bitmap indexes outperform B-tree indexes in the experiments.

Bitmap indexes are more suitable for data warehouses with large volumes of data.

The study offers practical insights for index selection in data warehouse design.

Abstract

With the development of decision systems and specially data warehouses, the visibility of the data warehouse design before its creation has become essential, and that because of data warehouse importance as considered as the unique data source giving meaning to the decision. In a decision system the proper functioning of a data warehouse resides in the smooth running of the middleware tools ETC step one hand, and the restitution step through the data mining, reporting solutions, dashboards... etc other. The large volume of data that passes through these stages require an optimal design for a highly efficient decision system, without disregarding the choice of technologies that are introduced for the data warehouse implementation such as: database management system, the type of server operating systems, physical server architecture (64-bit, for example) that can be a benefit performance of this system. The designer of the data warehouse should consider the effectiveness of data query, this depends on the selection of relevant indexes and their combination with the materialized views, note that the index selection is a NPcomplete problem, because the number of indexes is exponential in the total number of attributes in the database, So, it is necessary to provide, while the data warehouse design, the suitable type of index for this data warehouse. This paper presents a comparative study between the index B-tree type and type Bitmap, their advantages and disadvantages, with a real experiment showing that its index of type Bitmap more advantageous than the index B-tree type.