【作者】 吕桦；

【导师】 钟诚；

【作者基本信息】 广西大学 ， 计算机应用技术， 2006， 硕士

【摘要】 计算机和高速网络的日渐普及使得聚集地理分布的资源以形成计算网格成为可能。由于网格的巨大计算潜力，网格及相关技术得到了飞速发展。但是如何将这一潜力无缝地提供给用户是一个富有挑战性的问题。其中的一个主要困难在于如何在网格这样的大规模系统中调度一个大型的应用。网格虽然在本质上属于一种形式的分布式计算系统，但是由于网格资源在广域上分布、本质上异构、归属不同的组织和个人拥有、不同的存取和花费模式、负载和可用性动态变化等，使得传统分布式计算环境中的任务调度技术并不能直接的应用于网格领域。 随着网格的发展，各种各样的资源和应用加入网格，不同的资源提供不同级别的服务质量(QoS)，不同的应用任务要求不同级别的QoS。在这种情况下，QoS成为网格任务调度过程中必须要考虑的一个重要因素。而已有的启发式任务调度算法大多并没有考虑QoS的影响，这就可能使得那些低QoS的任务占用高QoS的资源迫使高QoS的任务等待而造成资源的浪费。据此，有人在传统的Min-Min启发式调度方法的基础上提出了一种新的自适应的QoS指导的Min-Min启发式调度方法。该启发式调度方法利用最新的预测技术来解决网格的动态性问题，同时在调度过程中考虑了QoS的影响。但是，上述算法在更多还原

【Abstract】 With the wide availability of powerful computers and fast networks, it is possible to aggregate geographically distributed resources to make a computational grid. For its grand computing potential, Grid and its relative technology are experiencing a fast development. However, how to provide seamlessly the potential to the users is still a challenging question. One of the major difficulties is how to schedule a large application in such a large scale environment like Grid. Though in essence a distributed computing system, the Grid has its unique characteristics: the resources are distributed in wide area, heterogeneous in essence, owned by different institutions and individuals, subject to various access and cost modes, time-varying in load and availability and so on. Hence the task scheduling techniques in traditional distributed computing environment can’t be applied in Grid directly.With the proliferation of the Grid, more and more resources and applications are joining the Grid, and different resources provide different levels of QoS and different applications require different levels of QoS. QoS is becoming an important factor which must be considered in the process of task scheduling. According to the fact that most of known task更多还原