失效恢复机制下的网格任务冗余调度优化

Optimal Redundant Scheduling of Grid Tasks Based on Fault Recovery

网格技术是目前学术界和工业界解决计算密集型问题的一种重要工具。由于网格系统的复杂性,网格在可靠性方面仍面临着诸多问题。针对目前网格服务可靠性低的问题,引入本地失效恢复机制,并允许资源自行调节网格任务生存时间以及失效恢复次数,从而建立更加符合实际的网格服务可靠性模型。在建模中,采取网格任务冗余调度方式,以进一步提高网格服务可靠性。基于建立的考虑失效恢复机制的网格服务可靠性模型,建立费用约束下的资源冗余调度优化模型,以获得网格服务可靠性最大的任务调度策略。针对该NP问题,采用遗传算法对该优化问题进行求解,并在求解过程中设计专门的修正算子校正不可行个体,以保障算法的正常运行。算例分析验证了算法的有效性。

Grid technology is an important tool to solve computation-intensive problems in current academic and industrial circles.Due to the complexity of grid system,there are a lot of problems unsolved in grid reliability.To cope with the low level of grid service reliability,a fault recovery mechanism in grid resources is introduced and a more practical grid service reliability model is proposed,in which the constraints on the life numbers of subtasks and on the numbers of recoveries performed can be imposed freely by resource owners.In order to further improve grid service reliability,a redundant scheduling of grid tasks is used.Based on the proposed grid service reliability model considering fault recovery,an optimization model with a cost constraint for redundant scheduling problem is presented to maximize the grid service reliability.A genetic algorithm is developed to solve it and some special repair operators are designed to adjust the infeasible solutions of the chromosomes,which can ensure normal algorithm operation.A numerical example is given to show the efficiency of the genetic algorithm.