Abstract: Based on the capacity job scheduling algorithm, this paper proposes a Map Reduce job scheduling algorithm based on Simulated Annealing(SA). It uses the SA algorithm with remember function to choose the best job, and avoids losing into local optimal solution. Experimental results on Hadoop platform show that the algorithm can reduce the total time of the jobs and the waiting time of each job, and it also has high job scheduling efficiency and satisfaction of customs.

Key words: cloud computing, job scheduling, Hadoop platform, Simulated Annealing(SA), Map Reduce model, local optimum

摘要： 在计算能力作业调度算法的基础上，提出一种基于模拟退火的Map Reduce作业调度算法。利用带记忆功能的模拟退火算法选择最优作业，从而避免陷入局部最优解。在Hadoop平台上的实验结果表明，该算法能减少所有作业的运行时间以及每个作业的等待响应时间，具有较高的作业调度效率及用户满意度。

关键词: 云计算, 作业调度, Hadoop平台, 模拟退火, Map Reduce模型, 局部最优

CLC Number: 

• TP302.7