Abstract:
Effective data management is an important issue with more and more scientific applications manipulating massive distributed secondary-storage, this paper presents a novel application development architecture which includes a Meta Data Management System(MDMS) and Storage Systems(SS) that provide a platform for easy-to-use and automatic storage access optimizations. It adopts a novel multi-storage resource architecture that can satisfy both performance and storage capacity requirements, and employs an adaptive I/O optimization schemes.
Key words:
high performance computing,
massive storage,
parallel file system,
meta data management,
access optimization
摘要: 在高性能计算程序对海量分布存储数据的操控中,有效的数据管理很重要。该文提出一个新的高性能分布计算的数据管理与优化系统,它包含一个元数据管理系统和存储系统,提供一个容易使用且能自动进行存储访问优化的平台。该平台采用的多存储资源体系结构能够满足性能和存储容量需求,并能自适应地利用当前的I/O优化方法。
关键词:
高性能计算,
海量存储,
并行文件系统,
元数据管理,
访问优化
CLC Number:
WEI Wen-guo; XIE Gui-yuan. Data Management and Optimization in High Performance Distributed Computing[J]. Computer Engineering, 2008, 34(13): 64-66.
魏文国;谢桂园. 高性能分布计算的数据管理与优化[J]. 计算机工程, 2008, 34(13): 64-66.