Hadoop平台中作业调度算法分析与改进研究

Analysis and improvement of job scheduling algorithm in Hadoop platform

针对传统云计算作业调度中简单优先级设置的不足,首先对作业调度进行研究并提出了基于用户服务等级协议(SLA)的作业分级机制。该设计基于Qo S约束的多优先级作业调度算法,在算法中利用云用户作业中的偏好程度来设计优先级值计算函数Priority,在作业调度时可以使较高等级的云用户和具有较高优先级的作业优先得到任务的执行,通过上述方式可以较好地保证云计算环境下的服务质量。

In view of the shortcoming of simple priority setting in the traditional cloud computing job scheduling,this paper studies the job grading mechanism firstly and puts forward the job grading mechanism based on user's service level agreement(SLA).The priority job scheduling algorithm is designed based on Qo S constraints.In the algorithm the cloud users' preference can be used to design the priority value to calculate function priority.When scheduling jobs,tasks from higher lever of could users and higher priority jobs can be performed finstly.This mentioned methods can better guarantee the quality of service in cloud computing environment.