Research On Energy-Saving Algorithm In Cloud Datacenter

Cloud computing is an emerging business computing model, computing tasks are distributed into a resource pool consist of large number of computers, user can get the computing, storage space, and information services according to their needs. The Internet has entered the cloud computing era, architecture and industrial mode of the IT industry will be completely changed, the urgent need to improve resource utilization of datacenter, prompting people to seek new ways to build the next-generation datacenter, traditional datacenter will gradually be replaced by cloud computing datacenter.Cloud datacenter can achieve efficient computing and managing through large-scale integration of resources, and provide infrastructure and application services to users through supply chain. With the support of technology like virtualization, task scheduling problem in cloud datacenter is redefined. How to make full use of the advantages of the scalability and flexibility of cloud computing technology to improve resource utilization and reduce energy consumption of cloud datacenter is one of the major issues in research on cloud computing technology.Therefore, designing energy-efficient scheduling algorithm for different scenarios in green cloud datacenter is of great importance. Specifically, the main contents and innovation of this research are described as follows:First of all, this paper introduce the concept of cloud computing and cloud data, and analyze the characteristics of the traditional datacenter and cloud datacenters. Through the analysis of the various factors that affect the energy consumption of cloud datacenter, energy consumption of servers is determined to be the main object of the energy consumption optimization, a mathematical model of server energy consumption is established.Secondly, we study the problem of energy-efficient scheduling for real-time tasks. A task scheduling algorithm based on the slack time of tasks is designed. Three components of the algorithm are the distribution of real-time tasks, dynamically expanding virtual machine resource and integration of virtual machine resource. By computing the slack time of tasks, tasks can be completed within the deadline to ensure the timeliness of the task. With the use of a multi-threshold-based virtual machine integration strategy, the system load is balanced and energy consumption of the system to complete the task set is reduced. Again, for the scheduling problem of DAG workflow under the cloud computing environment, this paper presents an energy-efficient scheduling algorithm based on DVFS technology. The algorithm is divided into two stages:In the pre-scheduling stage of the algorithm, the makespan of the task set is optimized by calculating the task priority.In the stage of energy-efficient scheduling, by delaying the execution of some task within the deadline, we can make better use of the idle time of the processor.Finally, the results of simulation experiment on simulation platform CloudSim comparing with two other scheduling algorithms prove that our algorithms are able to guarantee the timeliness of the tasks, and can efficiently reduce energy consumption of the system.