A novel cooperative resource provisioning strategy for Multi-Cloud load balancing

Abstract

The paradigm of cloud computing has heralded a new avenue of computing, offering benefits of increased data accessibility with low cost. Continuous Writing Applications (CWA) (e.g., augmented online services for Health Care) have specific requirements on data storage, computation and bandwidth, thus are cost-sensitive with limited budgets and time. Herein, we propose an architecture of multi-cloud service provider (CSP) or “Multi-Cloud” to provide services to CWA, and design a novel resource scheduling algorithm to minimize the system cost. The system models of classic CWAs to tackle the resource requirements of users on MCP are exploited. The study can help to understand the characteristics of different resources and conclude Multi-Cloud being the most attractive to many CWA implementations. Interconnections of multiple CSPs and their load paths (i.e., data passing through possible interconnections) are introduced. We then formulate the problem and present optimal user scheduling based on Minimum First Derivative Length (MFDL) of system load paths. Theoretical analysis demonstrated that the solutions with minimized costs can be achieved by the proposed algorithm, termed “Optimal user Scheduling” for Multi-Cloud (OSMC). Through rigorous simulations regarding different influencing factors, the proposed strategy has proven to be scalable, flexible, and efficient in many practical scenarios.