摘要
提出一种基于资源类型的非集中式网格资源发现方法 ,将注册有同类资源的网格信息节点组织在一起形成社区 ,资源发现请求的转发以及资源信息的扩散都被限制在相应的社区内 ,从而改善了资源发现的性能 社区的自组织和社区间的高效导航通过一个基于分布式Hash表的bootstrap网络实现 ,并提出了流言传播的机制来改善bootstrap网络中的负载均衡 在社区内部的搜索中引入了基于贪婪搜索的资源定位机制 ,只将少数较好的结果返回给用户 ,节省了用户比较选择的时间
A decentralized grid resource discovery scheme based on resource classification is studied in this paper, in which information nodes with the same type of registered resources are organized into communities, and a DHT based bootstrap network is constructed to navigate between different communities efficiently.Periodical topology maintenance communications in bootstrap network are used to piggyback and disseminate popular data in bootstrap network to achieve better load balance.A greedy search based resource location method is also introduced to identify qualified resource providers with good qualities inside communities, which improves scheduling efficiency.The performance improvement is evaluated by simulation under different cases, and overhead is also studied.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2004年第12期2156-2163,共8页
Journal of Computer Research and Development
基金
国家自然科学基金项目(70271004)