Hybrid Cloud Computing: Economy, Scalability and Responsiveness Optimization

Abstract

Hybrid cloud computing gains interest from communities as it supports risk mitigation, business partnership, quality-of-service (QoS) improvement and accesses to uniquely-offered services. Since the providers of a hybrid cloud potentially offer different QoS levels, this sets the new condition of cloud data transfer optimization to reduce public cloud data-out expenses, to improve cloud network scalability and to lower cloud service access latencies. This paper presents an intelligent cloud cache replacement policy, i-Cloud, as the core mechanism of client-side shared cloud cache. Trace-driven simulations have showed that i-Cloud is capable of addressing nonuniform QoS levels by delivering stable performances that outperformed three well-known cache replacement policies in all studied performance metrics against all experimented workloads. The results have also indicated that taking data-out charge rate nonuniformity into cache replacement decisions improved caching performances in all metrics. Furthermore, i-Cloud not only attained optimal efficiencies in all of the performance metrics simultaneously but also performed well for longer runs than its training durations.