科学工作流技术研究综述  

A Review on Scientific Workflows

作　　者：张卫民[1] 刘灿灿[1] 骆志刚[1] 

机构地区：[1]国防科技大学计算机学院,湖南长沙410073

出　　处：《国防科技大学学报》2011年第3期56-65,共10页Journal of National University of Defense Technology

基　　金：国家863高技术资助项目(2006AA01A123);国家自然科学基金资助项目(609030420)

摘　　要：科学工作流管理系统(Scientific Workflow Management Systems,SWfMS)为科学家提供一个管理大规模复杂科学计算的有效平台,通过管理复杂应用程序对海量科学数据进行管理、分析、仿真和可视化以帮助科学家进行科学发现。本文首先对科学工作流(Scientific Workflow,SWF)的关键技术和研究现状进行全面综述,包括SWF模型、表示、语言、组合、验证、调度、数据来源管理以及系统的容错性与安全性等;然后对近两年的最新研究进行分析,并在此基础上指出该领域目前存在的不足和未来发展方向;最后对我国目前的研究现状进行分析并给出建议。Scientific Workflow Management Systems （SWIMS） ｜ms become an effective platform for representing and managing largescale complex scientific computing processes, which glue together a large number of complex applications for data management, analysis, simulation, and visualization, and then assist the scientists for scientific discovery. In this paper, the current studies of Scientific Worldlow （SWF） are reviewed firstly, including SWF models, presentations, languages, composition, validation, scheduling and provenance, as well as fault tolerance and system security. Then, the latest studies in recent years are also discussed. Based on the analysis of the key technologies and the latest development, the drawbacks in this domain are pointed out and some suggestions towards the future develooment are made. Finally, the recent research status quo in China is presented and some suggestions are given.

关 键 词：科学计算 科学工作流(SWF) 流程组合 调度 容错 来源管理 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]