提出一种基于随机服务模型的可靠性量化计算方法,该策略采用随机服务系统描述网格资源的预留服务模型,通过分析预留请求的统计特征来评估接纳当前预留请求不发生违约的概率,然后依据备选资源集合中的违约概率来进行预留请求接纳和任务资源选择,从而提高资源预留时的可靠性。实验表明,与传统的预留策略相比,该策略在提高资源利用率、降低违约率以及对动态负载压力变化情况下的反应方面,表现出更好的有效性、适应性。
A stochastic model was used to describe reservation service of grid resources and an approach calculating re-liability-guarantee expressed as probability guarantee success rate was proposed for reservation requests.By the approach,a reliability-enhenced reservation strategy was proposed.Experimental results show that the proposed strategy can sig-nificantly improve resource utilization.Also,it is able to reduce the jobs' violation rate comparing to conventional reser-vation policy,with respect to adaptability to dynamic grid environment.