Optimizing Data Replication in Cloud Computing Using Firefly-Based Algorithm for Selection and Placement

Abstract

The rapid adoption of cloud computing has driven extensive research into data replication methods and their practical applications. Data replication is a vital process in cloud systems, ensuring data availability, improving performance, and maintaining system stability. This is especially crucial for data-intensive applications that require the distribution and sharing of large volumes of information across geographically dispersed centers. However, managing this process presented significant challenges. As the number of data replicas increases and they are distributed across multiple locations, the associated costs and complexity of maintaining system usability, performance, and stability also rise. In this study, we initially randomized the distribution of data replication files across the cloud infrastructure to simulate a realistic scenario where data already exists within the system before the application of replication algorithms. This approach allowed the algorithms to optimize the replication process based on the initial data distribution and adapt to the evolving demands of incoming workloads. To address the challenges of dynamic data replication in cloud environments, this paper introduced two algorithms: the Firefly Optimization Algorithm for Data Replica Selection (FFO-S) and the Firefly Optimization Algorithm for Replica Placement (FFO-P). A detailed simulation study was performed using the CloudSim platform to assess the effectiveness of the proposed FFO-S and FFO-P algorithms. The simulation environment was designed to closely emulate real-world cloud infrastructures, ensuring the practical applicability of the results.