云计算环境下面向科学工作流应用的弹性调度机制研究

项目名称： 云计算环境下面向科学工作流应用的弹性调度机制研究

项目编号： No.61202449

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 东方

作者单位： 东南大学

项目金额： 25万元

中文摘要： 当前工作流模式已广泛存在于不同领域的科学计算当中，其具有应用规模大、逻辑关系复杂等特点。近年来，为满足日益增长的科学计算需求，利用云环境实现科学工作流应用的高效运行逐渐得到关注。但现有工作并未考虑云计算特点，忽略了由虚拟机间竞争资源而导致的性能波动，且极少针对系统能耗进行优化。针对以上问题，本项目拟从分析云计算的优势与特点以及复杂工作流的调度模式入手，在确保应用执行性能的前提下以尽量优化系统能耗为目标，研究面向科学工作流应用的弹性调度机制。通过建立云计算环境下的任务性能与资源能耗模型，实现对虚拟机性能波动及系统能耗的感知；通过结合复杂工作流应用的执行特点，提出以性能约束与能耗优化为目标的弹性调度算法及任务动态迁移策略，有效保证应用的执行性能并尽量降低系统的整体能耗。本项目的研究工作将为云环境中复杂科学工作流应用高效率、低能耗地调度与执行提供切实有效的解决方案，具有重要的理论意义和应用价值。

中文关键词： 云计算；科学工作流；能耗；弹性调度；虚拟化

英文摘要： Nowadays, workflow applications, which consist of numerous tasks and complex logical relationship, are widespread in the area of scientific computing. Recently, with the increasing demand of scientific computing, more and more researchers foucs on utilizing cloud environment to execute scientific workflow applications effecitively. However, most of the existing works failed to take into account the characteristics of cloud computing. It may lead to the performance fluctuations caused by resource sharing among virtual machines and the large energy consumption caused by neglecting the energy optimization. To tackle these problems, this project tries to utilize the characteristics of cloud computing to develop an elastic scheduling mechanism for scientific workflow applications, in order to optimize system energy consumption when the performance has been guaranteed. On the first stage of this project, the model of task execution performance and energy consumption will be developed to perceive the performance fluctuations and energy change of system. Then, for the sake of performance guarantee and energy saving, the elastic workflow scheduling algorithm and dynamic task migration strategy will be proposed respectively. In summary, these works will render several effective and valuebale solutions for accomplishing co

英文关键词： Cloud Computing；Scientific Workflow；Energy Consumpution；Elastic Scheduling；Virtualization