提高可扩展并行机群并行I/O效率的一个方法

A NEW METHOD TO IMPROVE THE I/O EFFICIENCY ON NETWORK OF WORKSTATIONS

随着CPU性能的高速提升，系统I／O能力的不足越来越明显地成为提高NOW系统整体性能的瓶颈．在分析现有基于NOW系统的并行I／O算法的基础上，通过理论推导，给出了一种寻求计算进程与计算结点之间最佳映射的方法．该方法可以在数据重分配时，使各计算结点间的通信量最小，从而达到提高系统并行I／O效率的目的．

With the rapid improvement of the performance of CPU, the weakness of system I/Ois increasingly becoming the bottleneck to increase the whole capability of the NOW system.After generally analyzing the present parallel I/O algorithms, an optimum mapping strategy is given, which maps the local processes to physical processors. Using this method the communications can be decreased maximally when relocating data, so that the efficiency of thesystem parallel I/O can be increased.