Network-Aware Optimization for Efficient Data Placement in Geo-Distributed Cloud Systems

Abstract Explosive growth in IoT and cloud infrastructures presents significant challenges for data placement across geographically dispersed data centers. Existing placement strategies improve performance but often fall short of substantially reducing transmission latencies and ensuring load balance across these centers. This paper proposes the Optimal Data Placement Method (ODPM) for effective data placement strategies in geographically distributed cloud environments. ODPM has two parts; the first one is an integration of Vogel’s Approximation Method (VAM) and the Modified Distribution (MODI) technique to deal with cost reduction, load balance distribution, and capacity constraints in distributed cloud data centers. The VAM ensures an efficient starting point for solving the data transportation problem, while MODI method works on optimizing the solutions provided by VAM. The second part uses Floyd’s shortest path algorithm to find the shortest distance between IoT devices and cloud storage to minimize data transmission time and cost. The combined approach focuses on three aspects, the data transmission time, computation time, and load balance between cloud data centers. The experimental results demonstrate that ODPM reduces the data transmission time by 12% over current techniques such as Spectral Clustering on Hypergraphs (SpeCH) and data placement using Lagrangian relaxation (DPLR), and an 8% reduction in computation time for optimal paths calculation while 15% enhancement in load balancing. Integrating MODI and Floyd’s algorithm, ODPM constitutes an efficient and realistic approach to big data allocation in contemporary cloud environments.