计算网格中一种可用性增强的作业执行模型

如何保障用户QoS是动态可用性资源组成的网格环境中面临的一个新问题。针对网格资源的可用性特点，提出了可用性增强的作业执行服务(Availability-enhanced Job Execution Service，AJES)的概念及其确定方法，基于该概念提出了可用性增强的作业执行模型(Availability-enhanced Model for Job Execution，AMJE)，通过将作业调度到AJES包含的1个或多个网格节点中的资源上冗余执行以增强作业所需的可用性。将模型中的各方根据其效益描述为一个博弈过程，求得了在资源收益和用户QoS之间平衡的AJES可用性值。实验结果表明本文提出的作业服务模型能显著提高资源收益和增强用户QoS的保障水平；并且，对于可用性动态变化的网格环境具有更好的适应性。

An Availability-enhanced Model for Job Execution in Computational Grids

User’s QoS guarantee is a new challenge in Grid integrated with dynamic available resource. On the basis of quantifying resource availability, a definition called Availability-enhanced Job Execution Service (AJES) and its constructing method is proposed. Then, an Availability-enhanced Model for Job Execution (AMJE) is derived from AJES. Availability required for user’s job is enhanced through scheduling task to one or more Grid sites in AJES redundantly. We describe different parties in AMJE on behalf of their profits as a Game, and get a balance value for AJES availability between resource provider’s profits and user’s QoS. We conduct extensive experiments, and the results show that AMJE not only can increase resource provider’s revenue and improve QoS level for end users, but also exhibits good adaptness to dynamic Grid environment.