基于计算资源运行时剩余能力评估优化云平台
发布时间：2018-07-26 07:03
【摘要】：云平台资源管理中存在资源供给与需求不匹配的问题,导致平台性能受到严重影响.针对此问题,基于相似任务建立运行时计算资源剩余能力评估模型,该模型利用云计算负载中相似任务执行逻辑相同的特点,使用相似任务代替测试程序量化资源剩余能力,避免了执行测试程序的计算资源代价;依据该模型提出了一种运行时云计算资源剩余能力分类评估方法 RCE(resource capacity evaluation),该方法综合各方面因素评估运行时资源剩余能力,具有运行时代价低、评估结果准确且有时效性的特点.将RCE评估结果应用在若干算法中,以提高云平台资源供给与需求的匹配程度并优化云平台各方面性能;在独享环境和真实云环境中验证了RCE方法和基于RCE的算法,实验结果表明:RCE评估结果及时反映了计算资源能力变化,为算法和平台的优化提供了有力支持,基于RCE优化的算法解决了云计算资源管理中资源供给与需求不匹配问题并大幅提高云计算平台性能.
[Abstract]:The problem of resource supply and demand mismatch in cloud platform resource management results in a serious impact on platform performance. To solve this problem, a run-time computing resource residual capability evaluation model is established based on similar tasks. The model utilizes the same logic of similar tasks in cloud computing load, and uses similar tasks instead of testing programs to quantify resource residual ability. Based on this model, a method of classifying and evaluating the residual capability of cloud computing resources at run time, RCE (resource capacity evaluation), is proposed, which synthesizes various factors to evaluate the capability of resource surplus at run time. It has the characteristics of low operating time price, accurate evaluation results and timeliness. The RCE evaluation results are applied to several algorithms to improve the matching degree between resource supply and demand of cloud platform and optimize the performance of cloud platform in all aspects, and verify the RCE method and the algorithm based on RCE in the exclusive environment and real cloud environment. The results of the experiment show that the results of the RCE evaluation reflect the change of computing resources in time, and provide strong support for the optimization of algorithms and platforms. The algorithm based on RCE solves the problem of resource supply and demand mismatch in cloud computing resource management and greatly improves the performance of cloud computing platform.
【作者单位】： 西安交通大学电子与信息工程学院;
【基金】：国家重点研发计划项目(2016YFB0200902) 国家自然科学基金项目(61572394)~~
【分类号】：TP301.6
，
本文编号：2145228
[Abstract]:The problem of resource supply and demand mismatch in cloud platform resource management results in a serious impact on platform performance. To solve this problem, a run-time computing resource residual capability evaluation model is established based on similar tasks. The model utilizes the same logic of similar tasks in cloud computing load, and uses similar tasks instead of testing programs to quantify resource residual ability. Based on this model, a method of classifying and evaluating the residual capability of cloud computing resources at run time, RCE (resource capacity evaluation), is proposed, which synthesizes various factors to evaluate the capability of resource surplus at run time. It has the characteristics of low operating time price, accurate evaluation results and timeliness. The RCE evaluation results are applied to several algorithms to improve the matching degree between resource supply and demand of cloud platform and optimize the performance of cloud platform in all aspects, and verify the RCE method and the algorithm based on RCE in the exclusive environment and real cloud environment. The results of the experiment show that the results of the RCE evaluation reflect the change of computing resources in time, and provide strong support for the optimization of algorithms and platforms. The algorithm based on RCE solves the problem of resource supply and demand mismatch in cloud computing resource management and greatly improves the performance of cloud computing platform.
【作者单位】： 西安交通大学电子与信息工程学院;
【基金】：国家重点研发计划项目(2016YFB0200902) 国家自然科学基金项目(61572394)~~
【分类号】：TP301.6
，
本文编号：2145228
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2145228.html
