一种基于QoS的自适应网格失效检测器

失效检测器是构建可靠的网格计算环境所必需的基础组件之一.由于网格中存在大量对失效检测有着不同QoS需求的分布式应用,对于一个网格失效检测器来说,为保持其有效性和可扩展性,应该既能够准确提供应用程序所需的失效检测QoS,又能够避免为满足不同QoS而设计多套失效检测器所产生的多余负载.基于QoS基本评价指标,采用PULL模式主动检测策略实现了一种新的失效检测器--GA-FD(adaptive failure detector for grid),可以同时支持多个应用程序定量描述的QoS需求,不需要关于消息行为和时钟同步的任何假设.同时,证明了GA-FD在部分同步模型下可实现一个◇P类的失效检测器,并给出了相应的实验及数据.

An Adaptive Failure Detector for Grid Based on QoS

Task scheduling is a fundamental issue in achieving high performance in grid computing systems. However, it is a big challenge for efficient scheduling algorithm design and implementation. In this paper, the problem of scheduling independent tasks on tree-based grid computing platforms, where resources have different speeds of computation and communication, is discussed. In contrast to minimizing the total execution time, which is NP-hard in most formulations, an integer linear programming model for this problem is presented. Using the model, the optimal scheduling scheme that determines the optimal number of tasks assigned to each computing node is obtained. With the optimal scheduling scheme, two demand-driven and dynamic heuristic algorithms for task allocation are proposed: OPCHATA (optimization-based priority-computation heuristic algorithm for task allocation) and OPBHATA(optimization-based priority-bandwidth heuristic algorithm for task allocation). The experimental results show that the proposed algorithms for the scheduling problem obtain better performance than other algorithms.