文章基本信息

标题：Efficient and Collective Global, Local Memory Management For High Performance Cluster Computing
本地全文：下载
作者：P. Sammulal, A. Vinaya Babu
期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2008
卷号：8
期号：4
页码：81-84
出版社：International Journal of Computer Science and Network Security
摘要：In Cluster Computing Environment the data latency time has significant impact on the performance when the data is accessed across clusters. Especially when executing data pertaining to satellite images for remote sensing or defense purposes, scientific or engineering applications. Designating a particular cluster for executing an application leaving behind the bandwidth aspects of accessing data across clusters can pose performance degradation. To overwhelm the stated problem we have proposed a new technique that buffers the data in Global Memory and Local Memory and regulate the need for communicating across clusters for data access. This Global Memory constitutes a permanent storage part and a temporary storage part, which are refreshed dynamically and at regular intervals based on the data access patterns. Local Memory regulates the Intra-Cluster communication for data access. Experimental results show performance improvement to considerable levels with the implementation of the concept, specifically when the cost of data access from other clusters is higher and is proportionate to the amount of data.
关键词：High Performance Cluster Computing, Job Scheduling, Memory Management