Abstract
The digital world with constantly evolution gives rise to the precious concept , "data" that is known as the black gold. In accordance with this evolution, database management systems, which play an important role in data valuation, are becoming an essential element of information systems and decision- making processes. With respect to the digital revolution, data is generated every second in a huge volume, by multiple sources and with different formats.Despite the fact that managing large and dispersed data is a problematic issue , we could not neglect the precious value that could potentially be gained through raw data exploration. This heterogeneity translates into the need for an integrated system to efficiently store, process and analyze the huge amount of scattered data. The phenomenon of huge data, known as big data, requires a decision-making system with an appropriate architecture that stores the heterogeneous data and supports the main characteristics of the big data environment, such as the data volume, the veracity, velocity and veracity. The data lake, which is a centralized storage system, is a good answer to these arising problems to receive raw data on a large scale in their native formats. Concerning this goal, it is clear that the infrastructure and architecture of the data lake have a significant impact on the profitability and functionality of the overall system. In this regard, the design and management of the data lake structure requires practical and innovative methods in order to achieve an integrated and optimal centralized repository. By considering the systematic structure of the data lake as well as the hierarchical architecture of the systems, a logistical vision could lead us to the defined objectives.The supply chain is a good example of logistics systems where hierarchical participants are coordi- nated within an integrated network in order to prepare a product or render services to targeted consumers. The logistics structure as well as the supply chain management strategies could be an innovative source of inspiration to design, manage and optimize a data management system based on a logistics vision.For this reason, the implementation of the analog method between systematic structures clarifies to what extent one could take advantage of management strategies derived from the supply chain to develop the architecture and performance of the data lake.In this thesis, we hypothesize that it is possible to describe a data lake and its functionality by comparing it to the logistical structure of a supply chain. On the basis of these objectives:First, we are interested in relying on several data lake architectures and verifying the effectiveness of these architectures on the performance of the data lake, in particular in relation to data governance and the quality of services.In a second step, we introduce the supply chain, supply chain management and the methods that are used frequently to optimize the supply chain. Furthermore we compare all the elements of this data lake logistics system and focus on their similar points in order to use the data lake supply chain management methods.Thirdly, we propose a new architecture for data lake based on supply chain definition thanks to the evolutionary process of modeling the structures of data lakes. We finish this work by optimizing the proposed data lake architecture with supply chain network design strategies and propose the methods to solve the defined mathematical optimization model.