Uma abordagem polystore baseada em Fog-Cloud

Abstract

There are several challenges encountered in integrating and manipulating data in IoT (Internet of Things) environments. The demand for processing, querying and integrating heterogeneous distributed data has grown, making it essential to design an environment capable of dealing with such criteria. Considering the complexity of these scenarios, the use of a Polystore architecture is proposed, which aims to facilitate data integration at runtime, allowing the union of different types of databases and the construction of integrated storage mechanisms and heterogeneous. The study focuses on solving the difficulty of processing, organizing, integrating and storing data in the cloud, taking into account the Cloud and Fog layers. The proposed approach aims to minimize the intensive use of resources in the cloud, promoting data processing increasingly closer to the edges of the network. To achieve this goal, it is essential to understand the interactions between the cloud, edge and fog layers, considering the performance, energy, latency and computational costs demands typical of IoT environments. The Polystore architecture is presented as a promising solution to address these challenges, enabling efficient data integration and more effective use of available resources. Since, by integrating heterogeneous bases, it is expected that, when searches are carried out, all infected data will be returned to the search in each of these bases. The focus of the work is to validate the use of the architecture in the Fog environment, providing a more efficient and economical approach to real-time data processing. Based on analysis and experimentation, we seek to answer specific questions related to the implementation and impacts of this architecture, promoting its use and effectiveness on a large scale. The proposal presented uses simulation, parallelism and Polystore to process data in a simulated environment with Cloud and Fog layers. It prepares fog nodes for simulation, forming a framework with dynamic resource allocation using MySQL, MongoDB and a heterogeneous database. The resulting system allows searching in heterogeneous databases and monitoring the use of computational resources.