当前位置: X-MOL 学术Int. J. Comput. Sci. Eng. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
A fault-tolerant hybrid resource allocation model for dynamic computational grid
Journal of Computational Science ( IF 3.1 ) Pub Date : 2020-11-26 , DOI: 10.1016/j.jocs.2020.101268
Sophiya Sheikh , A. Nagaraju , Mohammad Shahid

Effectual allocation of resources with fault tolerance is one of the important targets in any computational grid environment to accomplish the task execution on time. In this paper, a Fault Tolerant Hybrid Resource allocation Model (FTHRM) has been proposed to minimize the turnaround time for the batch of tasks while ensuring fault tolerance in a dynamic grid environment. The model uses the prior reservation mechanism to allocate the resources to tasks for guaranteed task execution. Further, resources are reserved in advance for time slots with resource configuration as required by the batch of tasks. Alternate resources are provided by the system in the case of resource failures. Here, the resource with the least previous workload and lowest execution time will be preferred over other resources in task allocation. The performance evaluation of FTHRM has been carried out by comparing it with Minimum Completion Time (MCT). The simulation study reveals that the proposed model viz. FTHRM outperforms MCT on considered parameters under study. Statistical testing has been conducted to test the level of significance using SPSS 20, also confirms the conclusions drawn in the simulation.



中文翻译:

动态计算网格的容错混合资源分配模型

具有容错能力的有效资源分配是任何计算网格环境中按时完成任务执行的重要目标之一。本文提出了一种容错混合资源分配模型(FTHRM),以在确保动态网格环境中的容错能力的同时,最大程度地减少任务批次的周转时间。该模型使用先验预留机制为任务分配资源,以确保任务执行。此外,根据一批任务的要求,为具有资源配置的时隙预先预留资源。如果发生资源故障,系统将提供备用资源。在此,任务分配中的其他工作将比之前的工作量最少,执行时间最少的资源优先。通过将FTHRM与最小完成时间(MCT)进行比较来进行性能评估。仿真研究表明,提出的模型是可行的。在研究的参数方面,FTHRM的性能优于MCT。已经使用SPSS 20进行了统计测试以检验显着性水平,也证实了仿真中得出的结论。

更新日期:2020-11-27
down
wechat
bug