Cloud Workflow Scheduling Optimization Based On Genetic Algorithm For Energy Consumption

Since Google launched the Google 101 plan in 2006,many well-known company companies launched their own cloud program one after another.For example,IBM Blue Cloud program,Amazon Elastic Compute Cloud program,China Mobile Bigcloud plan,Alibaba Group Cloud program and so on.Cloud computing has already become a hot topic,the global cloud services market will maintain a rapid growth.With the rapid development of cloud computing services,large-scale cloud computing data centers are established all over the world.However,high energy consumption and high pollution caused by data center is becoming a more and more serious problem,which have draw people's attention.Currently,study on energy consumption optimization and management are mainly focused on dynamically adjust the voltage/frequency of the service CPU,virtualization technology(including virtual machine migration,distribution,consolidation),reorganizing independent parallel tasks in order to reduce the number of servers needed(including concentration method,migration method,etc),shutting down the idle server or make it sleep to reduce energy consumption and so on.However,all of these methods rarely consider the scheduling optimization technology based on workflow.Currently,most cloud workflow execution/scheduling optimization methods including genetic algorithm usually concentrate on the optimization of the execution time or cost,and seldomly consider the energy consumption.A few cloud workflow execution/scheduling optimization methods based on DVFS technology consider the energy consumption,while the DVFS technology needs to adjust the operating voltage/frequency of the service CPU and repeatly shutdown/restart the server,which will affect the influence performance and abrase the server components and rise the purchasing and replacement costs of server.Consequently,there are some limits in practical application.Therefore,researching on the optimization of workflow scheduling for energy consumption in the cloud computing environment,improving the quality of cloud computing services,reducing the energy consumption of cloud computing data center,and creating a green cloud service system,have a great significance.In this paper,considering the situation that there are few effective cloud workflow scheduling optimization methods orienting energy consumption,in this paper we establish cloud workflow process model and resource model based on energy consumption,and propose energy consumption calculation method and cloud workflow scheduling optimization algorithm based on energy consumption.Our proposed algorithm is called cloud workflow group scheduling algorithm based on genetic algorithm(GWSGA algorithm).Compared with the existing scheduling optimization method which is concerned with the time or cost,the algorithm considers the energy consumption factor.The encoding style of GWSGA algorithm firstly divides the workflow tasks into groups,and then determines the number of groups of each task in the workflow,and always put the tasks with small group number at the front of tasks with bigger group number,to ensure the effectiveness of individual coding.According to the number of groups of workflow,task requirements and resource availability,the initial population can prevent infeasible solutions,and this is the biggest advantage of our GWSGA algorithm.Because of this advantage,our algorithm has a shorter searching time than other algorithms and thus be more effective and can be widely used in any other scientific workflow.Our simulation experiment results show that GWSGA algorithm proposed in this paper can not only ensure the time efficiency of the implementation of the cloud workflow,but also can effectively reduce the energy consumption of the host processing tasks.