基于应用偏好模糊聚类的网格资源选择

Grid resource selection based on preference-based fuzzy clustering

提出了在网格环境中为应用选择合适资源的策略,设计了相应的资源选择算法和资源选择架构,实现了该策略的网格应用。该策略将应用偏好定义为一个向量,向量的每个分量对应着应用对资源某方面性能的偏好,使用模糊聚类技术结合应用偏好将网格资源划分为不同性能的集合,然后应用资源选择算法评估各个集合的平均性能并从中选择合适资源。应用该策略选择资源,可以满足应用需求,缩小资源搜索空间,降低资源分配开销,也有助于实现网格环境的负载平衡。

This paper proposes a resource selection approach for application to locate proper grid resources, and a resource selection algorithm and a resource selection framework have been developed to realize grid application. It defines application preference as a vector with m elements, one element for one side of resource performance, and applies fuzzy clustering technology to create resource sets of different performances according to application preferences. Then it evaluates the performance of every resource set and selects suitable resources using the resource selection algorithm. Besides meeting application requirements, this approach can decrease the cost on resource selection by reducing the search space of available resources. Furthermore.load balance in grid environments can also be improved.