High Performance Parallel I/O

高性能并行 I/O

• 批准号: 0105565
• 负责人: Kartik Mohanram
• 金额: $ 25万
• 依托单位: William Marsh Rice University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-07-15 至 2007-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0105565&HistoricalAwards=false
• 关键词: Performance; Parallel

This project addresses the challenges of successfully deploying large-scale parallel I/O servers to meet the demands of modern data-intensive applications multimedia retrieval, We and database servers, visualization and graphics, and spatial and temporal databases. The goals are to develop scheduling and resource management algorithms for parallel I/O systems, and increase our understanding of the complex underlying resource tradeoffs. These include basic scheduling issues dealing with parallel I/O, including those related to prefetching and caching, on -line scheduling, fair servicing of multiple users and deadline-constrained real-time parallel I/O. Secondly the algorithms designed in this research will be directly applied to areas like multimedia systems, including variable-bit rate (VBR) video retrieval, and Web, database and application servers dealing with large numbers of concurrent, interacting I/Os.

该项目解决了成功部署大规模并行I/O服务器的挑战，以满足现代数据密集型应用程序的需求多媒体检索，We和数据库服务器，可视化和图形，以及时空数据库。目标是为并行I/O系统开发调度和资源管理算法，并增加我们对复杂的底层资源权衡的理解。这些问题包括处理并行I/O的基本调度问题，包括与预取和缓存、在线调度、多用户的公平服务和期限限制的实时并行I/O相关的问题。其次，本研究设计的算法将直接应用于多媒体系统，包括可变比特率（VBR）视频检索，以及处理大量并发交互I/ o的Web、数据库和应用服务器等领域。