基于QoS的网格资源聚类管理  

Clustering management of grid resource based on QoS

作　　者：马晨明[1] 黄德才[1] 吴天虹[1] 翁挺[1] 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310032

出　　处：《浙江工业大学学报》2009年第4期421-425,共5页Journal of Zhejiang University of Technology

基　　金：浙江省自然科学基金资助项目(Y107596)

摘　　要：资源管理是网格计算研究领域中的一项重要研究点,为满足当前服务网格对QoS的高要求,将服务质量有效地溶入到管理机制中,使资源能更好地在开放、异构和动态的不确定网格环境中得到有效利用具有重要的理论意义和现实意义.总结并提出了一种QoS的三层结构模型,依靠资源绑定的策略有效地解决了网格资源中管理难的特点,并利用改进后的DBSCAN聚类算法实现对资源属性的划分,将资源分成多个资源相似集合.仿真实验证实,该算法能有效实现资源的聚类.Resource management becomes a hot research field in grid computing area. In order to meet the current service grid for high QoS requirements, adding the service quality into management mechanism in an effective way has important theoretical and practical significance. It can let the resource work well in the open, heterogeneous and dynamic grid environment. The paper summarizes and proposes a kind of three-hierarchical structure of QoS. It relies a strategy of resources binding in order to effectively solve the difficult problem in grid resources management. Then taking advantage of the improved DBSCAN clustering algorithm, the division of attributes of resources is achieved. In this way, the resource can be divided into a number of similar resource sets. The experiment results show the algorithm can effectively cluster resources.

关 键 词：资源管理 QOS 服务 资源绑定 DBSCAN聚类 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]