基于动态冗余机制的数据并行型计算网格模型  

Data parallel computational grid model based dynamic redundancy

作　　者：陈庆奎[1] 那丽春[2] 

机构地区：[1]上海理工大学计算机工程学院,上海200093 [2]上海立信会计学院信息科学系,上海201620

出　　处：《通信学报》2005年第12期110-116,共7页Journal on Communications

基　　金：国家自然科学基金资助项目(60573108);上海自然科学基金资助项目(04ZR14100);上海局管科技发展重点基金资助项目(04JG05071);上海教委发展基金资助项目(04EB21)

摘　　要：在由多个普通计算机机群构成的网格环境下,为了解决大规模数据并行型计算加速比问题、容错问题和动态负载均衡问题,提出了一个基于动态冗余机制的数据并行型网格模型。给出了由多个计算机机群组成的数据并行型计算网格等一系列形式化定义。利用逻辑计算机机群、机群逻辑环、计算节点逻辑环和m-冗余策略,研究了动态冗余、动态负载均衡技术。根据这些技术,描述了网格环境下的数据并行型计算过程。理论分析和实践表明,该模型有效地解决了大规模数据并行型计算所需的负载均衡、容错和资源优化问题。该模型适合基于网格的数据并行型计算。For resolving the problem of speed up, dynamic loaded balance and fault-tolerance for large scale data parallel computation in grid that composed of many computer clusters, a data parallel computational grid model based dynamic redundancy was discussed. A series of formal definitions, such as the data parallel computational grid, were given. Using the mechanism of the logical computer cluster, the logical ring of logical computer clusters, the logical ring of computers and the strategy of m-redundancies, the techniques of dynamic redundancies and dynamic load balance were studied. The parallel computation algorithms of DG were described. The analysis and experiment results show that this model effectively resolves the problems of load balance, fault-tolerance and optimizing resource in grid. It can be fit for the data parallel computing based in grid.

关 键 词：网格 数据并行型计算 大规模并行算法 动态冗余 容错 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] TP301.6[自动化与计算机技术—计算机科学与技术]