| 基于SDN的改进LATE调度器 |
投稿时间：2019-11-27 修订日期：2019-12-02  点此下载全文 |
| 引用本文： |
| 摘要点击次数: 385 |
| 全文下载次数: 0 |
|
|
| 中文摘要:作为大数据集群Hadoop 的核心技术之一，作业调度算法对Hadoop集群作业的运行效率及资源的合理分配都产生着至关重要的影响。LATE调度算法虽然能够从某种程度对现有算法存在的一些问题进行缓解，然而其却存在分配节点执行落后任务的备份任务时的不足，基于此，我们提出一种将SDN结合在内的改进LATE调度算法，该算法将带宽这一重要因素融合在LATE调度算法内，通过与OpenFlow控制器协作获取网络相关信息，从而在分配任务时进行更为高效的调度。通过对改进算法的实验及性能分析，验证了该算法能提高MapReduce的执行效率。 |
| 中文关键词:SDN LATE HADOOP 调度算法 |
| |
| Improved LATE scheduling algorithm based on SDN |
|
|
| Abstract:As one of the core technologies of Hadoop in big data cluster, job scheduling algorithm plays an important role in the operation efficiency and resource allocation of Hadoop cluster. Although the late scheduling algorithm can alleviate some problems existing in the existing algorithm to some extent, it has some shortcomings in allocating nodes to perform backup tasks of backward tasks. Based on this, we propose an improved late scheduling algorithm which combines SDN and integrates the important factor of bandwidth into the late scheduling algorithm. To obtain the network related information through collaboration with OpenFlow control, we can efficiently assign tasks in an optimized way. The experiments and performance analysis for the algorithm show that it has significantly improved in the work of MapReduce. |
| keywords:Hadoop MadpReduce LATE SDN |
| 查看全文 查看/发表评论 下载pdf阅读器 |
