Abstract: Efficient task scheduling is crucial in dealing with business efficiently and cutting down the operating costs for cloud service providers.To improve theperformance of task scheduling in cloud environment,this paper proposed a new algorithm,namely greedy simulated annealing (G&SA).Firstly,it finds the local optimal solution by executing the greedy algorithm,which is used to initialize the current optimal solution of the G&SA algorithm and the initial solution of simulated annealing algorithm.Secondly,the current optimal solution is updated by simulated annealing algorithm.As a result,the experiment shows that the G&SA algorithm can achieve global convergence faster compared with the traditional task scheduling algorithm.In addition,the G&SA algorithm not only obtains more stable optimization results and improves the quality and efficiency of optimization,but also reduces the total task time costs.Average resource utilization rate of virtual machine is steady at 99% or more,and the load can be more balanced.

Key words: Cloud computing, G&SA algorithm, Load balancing, Task scheduling