CAJ | 학술논문

任务调度在云计算环境中发挥着重要作用。提出一种基于Kriging代理模型的动态云任务调度方法。通过对云任务在不同资源组合下的性能表现进行Kriging代理模型建模并优化，从而得到对应于该云任务的最优资源分配方案；利用云平台的API，可动态对该云任务实施资源调度。基于OpenStack开源云平台，对两个工程计算应用进行了任务调度性能测试，结果表明该方法可有效动态调整云任务中的资源配给，按需按优对平台中的云任务进行资源调度。
임무조도재운계산배경중발휘착중요작용。제출일충기우Kriging대리모형적동태운임무조도방법。통과대운임무재불동자원조합하적성능표현진행Kriging대리모형건모병우화，종이득도대응우해운임무적최우자원분배방안；이용운평태적API，가동태대해운임무실시자원조도。기우OpenStack개원운평태，대량개공정계산응용진행료임무조도성능측시，결과표명해방법가유효동태조정운임무중적자원배급，안수안우대평태중적운임무진행자원조도。
The task scheduling method plays an important role in the various cloud computing environments. A Kriging surrogate model based dynamic task scheduling method in cloud computing is proposed in this paper:with the help of Kriging surrogate model established on the specific resource allocations and the performances accordingly belonging to the cloud tasks, the optimal resource scheduling strategy could be obtained;resources for the cloud tasks will then be recon-figured using the cloud APIs. An OpenStack open source cloud platform with two engineering computing applications is set up for method testing, with the results demonstrating that the proposed method can effectively manage the resource configurations for the cloud tasks dynamically, and the tasks will be scheduled according to their optimal needed resources at the same time.