Research On The Energy-efficient Cloud Tasks Scheduling Strategy Based On Dynamic Genetic Algorithm

The rapid development of cloud computing has led to the establishment of large-scale data centers. Such data centers consume enormous amounts of electric energy, resulting in high operating cost and carbon dioxide emissions which has a lasting influence on global climates. With the constantly increasing of cloud computing scale, the problem of minimum consumption control of data centers is becoming increasingly apparent. It is of great economic and environmental importance to study on how to reduce the energy consumed by large-scale data centers.Task scheduling problem has been an everlasting research topic in the field of cloud computing which is the crucial point that determines the execution efficiency. Scheduling strategy handles the problem of how to assign tasks to the corresponding computing resources rationally and efficiently. The optimization object of a task scheduling strategy can be total execution time of all the tasks or the power consumption of data centers. In recent years, the usage of some intelligence methods has attracted broad attention. Genetic algorithm is one of the most famous artificial intelligence approach which is efficiently used to solve the cloud job scheduling problem.Combined with the analysis of cloud computing architecture, this paper focused on the data center level and proposed a energy-efficient dynamic genetic scheduling strategy which is based on traditional genetic algorithm. Two weighted optimization object of the proposed strategy are total execution time and total power consumption. By setting up weight coefficient, users can control the main focus of the scheduling strategy. The proposed scheduling strategy also fully considered the dynamic characteristics of cloud computing environment: using an improved iterative method to deal with the capacity changes of cloud environment, using improved genetic operators to ensure the effectiveness, proposed a reselection method to accelerate convergence speed so that the scheduling strategy can react quickly to capacity changes.At last, several cloud task scheduling scenes were constructed with the aid of Cloud Sim-3.0 simulation tools and the proposed scheduling strategy is compared with some famous scheduling strategy such as sequential scheduling strategy, greedy scheduling strategy and ant colony scheduling strategy. The contrast experiments not only validated the performance of reducing execution time and power consumption but also tested the reliability and the weight influence. The result showed that the efficiency of the proposed strategy in terms of dealing with the capacity changes and demonstrated the validity and advantages over reducing execution time and power consumption, especially in the condition of large input scale.