| 孙凌宇,冷明.基于不同分配策略的云计算任务调度性能比较与分析[J].井冈山大学自然版,2016,(1):62-68,74 |
| 基于不同分配策略的云计算任务调度性能比较与分析 |
| THE COMPARISON AND ANALYSIS OF TASK SCHEDULING PERFORMANCE IN CLOUD COMPUTING BASED ON DIFFERENT ALLOCATION STRATEGIES |
| 投稿时间：2015-10-28 修订日期：2015-12-03 |
| DOI：10.3969/j.issn.1674-8085.2016.01.013 |
| 中文关键词: 云计算 任务调度 分配策略 负载均衡 性能分析 |
| 英文关键词: cloud computing task scheduling allocation strategy load balancing performance analysis |
| 基金项目:国家自然科学基金项目(61363014,61163062);流域生态与地理环境监测国家测绘地理信息局重点实验室招标课题(WE2015012);江西省青年科学家培养对象计划 (20153BCB23003);江西省科技厅支撑项目(20132BBE50048);江西省自然科学基金项目(20132BAB201035);江西省教育厅科技计划项目(GJJ150779) |
|
| 摘要点击次数: 10066 |
| 全文下载次数: 17921 |
| 中文摘要: |
| 形式化描述了云计算环境下的负载均衡任务调度问题,借助动态规划方法形式化推导了最早完成时间的启发式优先分配策略,给出了基于先易后难优先分配策略、先难后易优先分配策略的启发式云计算任务调度算法。阐述了基于顺序调度策略、先易后难优先分配策略、先难后易优先分配策略等启发式任务调度算法和基于禁忌搜索策略、元胞演化策略等智能任务调度算法。针对不同分配策略的云计算任务调度进行性能比较与分析,提出了完成时间可改进百分比和资源负载平衡因子的调度性能评价指标,实验数据对比充分表明:与启发式调度算法相比,智能调度算法能减少任务执行时间,优化资源负载均衡性能。 |
| 英文摘要: |
| The formal description of load balancing task scheduling problem in cloud computing is presented. We make its formal derivation based on dynamic programming method and built the heuristic scheduling strategy of the earliest finish time (EFT) for task scheduling. Furthermore, we propose the priority-to-easy scheduling strategy and priority-to-difficult scheduling strategy for task scheduling in cloud computing based on the EFT strategy. We present the heuristic scheduling algorithm based on the sequential scheduling strategy, the priority-to-easy scheduling strategy and priority-to- difficult scheduling strategy. We also present the intelligence task scheduling algorithm based on the tabu search scheduling strategy and the cellular automata scheduling strategy. Then, we propose two evaluation factors of scheduling performance analysis, which are the improvement percent of the latest time and the load balancing factor. Finally, we carry out the comparative experiments of scheduling performance under the CloudSim simulation platform of cloud computing based on five allocation strategies, which are the sequential scheduling strategy, the priority-to-easy scheduling strategy, priority-to- difficult scheduling strategy, the tabu search scheduling strategy and the cellular automata scheduling strategy. The experiment and analysis show that intelligence scheduling strategy has better performance in comparison with the heuristic scheduling strategy in terms of the decreasing the task completing time and the improvement of resource load balancing. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
