A prediction based energy conserving resources allocation scheme for cloud computing

As new cloud computing technologies continue to be developed, the systems are more and more efficient. This has enriched the applications of cloud computing, ranging from industry, business, to scientific fields. Nowadays cloud computing has become one of the important research issues in the computing and computer network fields. A cloud computing system consists of several independent servers. By way of the virtualization technique, the system manages all of the computing resources efficiently to process each user demand. However, a great number of operating servers will bring considerable power consumption. Efficient resource allocation methods design is one of the important solution approaches to relieve this situation. A resource allocation method will generally allocate each arriving job to proper available computing resources (the virtual machines, VMs) based on the consideration of the related features (such as the job size, the arrival time, etc.) of jobs in the waiting queue. Although the future arrival jobs are unknown, they will significantly affect the resulting performance of the resource allocation. In this paper, we develop an Energy Conserving Resource Allocation Scheme with Prediction (ECRASP) for cloud computing systems. The prediction mechanism can predict the trend of arriving jobs (dense or sparse) in the near future and their related features, so as with help the system to make adequate decisions. Simulation results show that our proposed ECRASP method performs well compared to conventional resource allocation algorithms in the energy conserving comparisons.