基于多QoS需求驱动的网格资源调度研究

为解决网格用户多QoS需求的资源调度问题,引入了满意度函数模型和经典Min-Min算法。将众多网格QoS分为性能和信任两类,选取性能QoS中的优先级、时效性、精度性和信任QoS中的安全性、可靠性共五个指标,分别构建每一维QoS参数的满意度函数模型并形成QoS综合满意度函数模型,由此设计多QoS约束的网格资源调度(Q-Min-Min)算法,以期将Min-Min算法中按照期待执行时间(ETC)进行调度改为按照服务质量综合满意度(QSM)进行调度。仿真实验表明,改进的Q-Min-Min算法在任务的跨度和成本两项性能指标上均比Min-Min算法更具优势,取得了较为理想的结果,证明了基于多QoS需求驱动的网格资源调度的有效性。

Research on grid resource scheduling for multi-QoS demands

To solve the resource scheduling problem for multi-QoS demands, this paper adopted satisfaction function model and classical Min-Min algorithm. Firstly, it classified the numerous QoS parameters into two types which were performance QoS and trust QoS. Based on the priority, timeliness, precision of performance QoS and the security, reliability of trust QoS, it constructed five QoS satisfaction function models and formulated the overall satisfaction function model. Then it designed a multi-QoS constrained grid resource scheduling Q-Min-Min algorithm to compare with the classical Min-Min algorithm, in which the QSM replaced the ETC matrix. The simulation results show that Q-Min-Min algorithm gives more advantages over Min-Min algorithm in makespan and cost indexes, and prove the validity of grid resource scheduling based on multi-QoS demands.