Modeling and data warehousing have been considered, for more than one decade, as a new challenging research topic for which different approaches have been proposed. Nevertheless these proposals have focused on static aspects only. In practice, the evolution of the operational information system can lead to changes in its dependent multidimensional data warehouse (i.e. that this system feeds with data), and therefore may require the evolution of the data warehouse model. In this evolving context, the authors propose a model-driven based approach in order to automate the propagation of the evolutions occurred in the source database towards the multidimensional data warehouse. This approach is based on two evolution models, along with a set of transformation rules formalized in Query/View/Transformation. This paper describes this evolution approach for which we are developing a software prototype called DWE© (Data Warehouse Evolution).

Abstract:The Data Warehouse (DW) is characterized by complex architecture, specific modeling and design approaches. It integrates data issued from operational data sources in order to meet decision-makers' needs by providing answers for OLAP queries (On-Line Analytical Processing). In practice, both data source models and decision-makers' analytical requirements evolve over time and, therefore, lead to changes in the DW multidimensional model. In this evolving context, we have developed the DWE (Data Warehouse Evolution) framework. DWE automatically propagates the changes of the data source data-model on the DW data-model. This paper proposes a model-driven approach for extending DWE in order to consider a further related evolutionary aspect: The evolution of decision-makers' needs. It deals with the propagation of these evolutions on the DW multidimensional model. This approach relies on a classification of evolution scenarios and a set of transformation rules for the identification of evolution operations to apply on the DW.

No abstract

Abstract:A Data warehouse (DW) is characterized by a complex architecture, designed in order to integrate data derived from operational data sources (DS), hence providing advanced analytical tools of these data. The DW is highly dependent on its DS. Hence, evolutions of the DS schema need to be propagated to the DW schema and content. This paper presents a model-driven approach for the evolution of a multidimensional DW. It is based on two evolution models: a first evolution model for the DS and another for the DW. These two models concern the data structure aspects as well as the evolution operations. The transition between these two models is performed through specific transformation rules defined in QVT (Query\View\Transformation). INTRODUCTIONIn the data warehousing (DW) field, whatever the enterprise's philosophy falls into i) Bill Inmon's camp where the DW is one part of the overall BI system, or into ii) Ralph Kimball's camp where the DW is the conglomerate of all data marts within the enterprise, data issued from the operational systems are extracted, transformed, cleansed and finally loaded into the fact and dimension tables of the famous star schema which represents the keystone modeling diagram that has twofold objectives: first, it highlights the subject of analyses (i.e., fact representing the activity to be evaluated) and, secondly, it shows up the axes (i.e., dimensions) according to which the fact's data could be analyzed (Inmon, 2002). This strong dependency between the DW and the data source (DS) leads to a new evolution problem that addresses the impact of the DS schema evolution on the DW. In fact, the dynamic evolution of business processes within the enterprise can lead to another evolution of the DS schema. The associated DW cannot escape from this evolution which can simultaneously affect its schema, stored data, and also the ETL process (Extract-TransformLoad) (Vassiliadis, 2009). This paper treats this evolution problematic, it is organized as follows. In section 2, we overview researches related to the DW evolution problem.Section 3 proposes a model-driven approach for the propagation of DS schema changes towards the DW. Section 4 defines the evolution models of both the DS and the DW. Section 5 presents an example of transformation rules formalized in QVT (Query \View \Transformation); it is for the automatic passage between these two models. Finally, section 6 concludes the paper and enumerates our future perspectives. RELATED RESEARCHESThe DW evolution problem has been the subject of several research studies. It was treated from several points of views: Analytical need evolution, DS schema evolution, etc.Some researchers (Favre et al., 2007), (Benitez et al., 2004), (Blaschka et al., 1999 ) have limited their study to the DW evolution as a result of evolution of decision makers needs, without considering the case of the DS evolution. Other literature works (Bellahsene, 2002), (Wrembel and Bebel, 2007), (Solodovnikova, 2008) have examined the evolution of the source schema as well a...

No abstract