(IJCSIS) International Journal of Computer Science and Information Security, Vol. 11, No. 6, June 2013 Toward a New Approach for Modeling Dependability of Data Warehouse System Imane Hilal RITM Lab., CED Engineering Sciences ENSEM ESTC, Hassan II University Casablanca, Morocco Reda Filali Hilali RITM Lab., Computer Engineering Department Nadia Afifi RITM Lab., Computer Engineering Department ESTC, Hassan II University Casablanca, Morocco Mohammed Ouzzif RITM Lab., Computer Engineering Department ESTC, Hassan II University ESTC, Hassan II University Casablanca, Morocco Casablanca, Morocco Abstract—The sustainability of any Data Warehouse System (DWS) is closely correlated with user satisfaction. Therefore, analysts, designers and developers focused more on achieving all its functionality, without considering others kinds of requirement such as dependability’s aspects. Moreover, these latter are often considered as properties of the system that will must be checked and corrected once the project is completed. The practice of "fix it later" can cause the obsolescence of the entire Data Warehouse System. Therefore, it requires the adoption of a methodology that will ensure the integration of aspects of dependability since the early stages of project DWS. In this paper, we first define the concepts related to dependability of DWS. Then we present our approach inspired from the MDA (Model Driven Architecture) approach to model dependability’s aspects namely: availability, reliability, maintainability and security, taking into account their interaction. Keywords-component; Data Warehouse System; Model Driven Architecture ; Dependability; Availability; Reliability, Security, Maintainability. I. INTRODUCTION Data Warehouse Systems (DWS) are specially used by decision makers to analyze the status and the development of their organization [1], based on a large amount of enterprise information integrated from heterogeneous Data Sources [2]. This information is organized following a multidimensional structure to make exploration and analysis easier for users. The DWS’s architecture is composed from several layers: (i) Heterogeneous Data Sources (DS), (ii) ETL (Extraction/ Transformation/ Loading) process which extract and transform data from these DS, and load the information into DW ,(iii) Data Warehouse repository, (iv) Restitution Tools that analyze the data in OLAP way. The final goals to implement the DWS are: (i) To evaluate complex queries without causing severe impact on the sources; (ii) To increase data availability and decrease response-time for OLAP queries; (iii) To provide correct historical trends for state-oriented data; (iv) To protect and secure the crucial business information; (v) To back up modifications and insure evolution and maintenance. Those goals involve the guarantee of non-functional requirements such as availability, reliability, security, confidentiality, integrity and maintainability which are encompassed in the dependability’s attributes [3]. To meet those goals, we suggest an approach that spans over dependability’s attributes. Our contribution presents, on the one hand, the advantage of considering these attributes from the early stages of the DW project. On the other hand, it provides models to the designers and developers in order to realize these functionalities in respect to the Model Driving Architecture’s (MDA) principles. The remainder of this paper is organized as follows: section 2 will present the related work on the development of dependable DWS; section 3 will give an overview of DW, dependability aspects and MDA; section 4 will introduce our approach through which we develop our models for DWS dependability. An example of implementation is shown in section 5. Finally, in section 6 we present our conclusion and future work. II. RELATED WORK In DW’s literature, dependability’s attributes have been neglected or they have been presented as a second class type of requirements, and have also been considered as the other non-functional requirements. But the experiences show that capturing non-functional requirements without mapping them into the conceptual model may provoke loss of information [4]. Only few works have specifically addressed this issue. In 47 http://sites.google.com/site/ijcsis/ ISSN 1947-5500