【作者】 王一夫；

【导师】 陈松乔；

【作者基本信息】 中南大学 ， 计算机应用技术， 2008， 博士

【摘要】 网格是当前分布计算研究领域的热点之一,网格的关键技术是网格资源管理,而网格资源发现则是网格资源管理中的一个基本组成部分,它为网格资源调度寻找满足应用需求的各种资源。与传统分布式系统相比,网格中可集成的资源规模更大,种类更多,结构各异,且分属于不同组织,参与网格的各结点往往拥有不同利益和资源管理策略。随着网格的进一步发展,集中式的资源发现方式将会出现性能瓶颈,而且由于信任和安全方面的原因,集中式的方式难以保证获取足够的资源信息支持应用对资源发现的需求。因此,适应网格发展需求的资源发现方式不应该仅仅依赖集中或全局控制的非集中方式,而应能更好地适应大规模、动态的网格环境。论文研究非集中式网格资源发现和管理中的系统规模动态性与发现效率之间的矛盾及其关键技术,主要从以下几个方面展开了研究工作:1、路由算法和动态自配置特性是实现大规模基于关键字的资源发现方法的关键问题。本文提出具有组播集群的层次性系统模型,设计了混合式路由算法,充分利用物理网络组播特性,节省网络带宽;提出组播集群复制协议和基于关键字的组播树策略AMTP,分别处理节点或者链路失效导致的网络分割以及路由的重建。实验结果表明,这些算法和策略节省了网络带宽,提高了系统的性能,保证了系统的自配置特性。2、针对非结构化拓扑P2P系统的特点,提出基于结点分类的高效资源管理方法(CBS)及其“相似结点分类”拓扑优化技术,通过在各结点上维护压缩状态表指导搜索请求转发和通过搜索缓存提高搜索效率。能够在系统中形成主题相关的结点分类聚集,并提供部分远程结点的快捷连接。实验表明,与泛洪、随机转发和Neurogrid等方法相比,CBS方法通过维护少量状态信息在保持较低搜索延迟的同时,显著降低资源搜索的消息开销,从而提高非结构化拓扑P2P系统的可扩展性和性能。3、结合P2P技术提出了网格环境中的一种资源发现模型及其需要解决的关键问题和方法。设计了节点加入与离开算法和资源查找算法,加快了资源定位和对资源有效的利用。该模型和算法支持基于属性的动态分布式查找,具有很好的扩展性。4、服务质量(QoS)是衡量网格是否成功的重要因素,本文基于QoS指标,提出了一种网格资源功能与性能评估标准及其相应的指标优化机制。提出基于优化时间轴的资源预留机制,解决了“资源碎片”问题,提高了资源请求的接受率和有效资源的利用率。更多还原

【Abstract】 Grid is a hot topic in distributed computing nowdays. Resource discovery is a basic problem in Grid resource management, which concerns discovering resources in Grid to meet the requirement of applications. Compared with traditional distributed systems, Grid aims to integrate much more resources of varieties of Grid resources usually have different or even contradictory interests. With further development of Grid, centralized Grid resource discovery schemes have potential scalability and security problems. Therefore, in order to survive the dynamic and larger-scale Grid environment, resource discovery should be decentralized and should not rely on centralized coordination and control.Focusing on scalable and efficient decentralized resource discovery in Grid environment, this paper makes following contributions:1. Key-Based publish/subscribe systems have recently received an increasing attention. Efficient routing algorithms and self-configuration are two key issues in the area of large-scale content-based publish/subscribe systems. This paper first proposes a hierarchical system model with multicast clustering. Then a hybrid routing algorithm is presented, which can fully exploit multicast in order to reduce the used network bandwidth. Moreover, a multicast clustering replication protocol and a content-based multicast tree protocol are presented for coping with the node or link failures and rebuilding the event dispatcher trees. Experimental results reveal that the system has better routing efficiency and lower cost, and guarantees the self-configuration characteristic.2. This thesis proposes CBS, a novel class-based search approach for unstructured P2P systems. CBS adopts similarity class topology optimization technique to form classes between peers with related topics and provide shortcut connections to some remote pees. CBS maintains a compressed state table on each peer to guide resource search, and adopts search cache to utilize the locality of searches. Analysis and simulations show that compared with flooding, random forwarding, Neurogird or other similar approaches, CBS can reduce the search message cost remarkably while keeping low search latency by maintaining a little state information, and thus it can improve the scalability and performance of unstructured p2p systems.3. The article makes up a module of resource discovering in the Grid environment, and brings out the key problems and methods in this module. It also delves in the algorithm of the addition and leave of node and the algorithm of resource researching and quickens position the resource and makes use of the resource effectively. This module supports Property-Based dynamical distributed search, and it also can be extended and optimized easily.4. QoS(Quality of Service) is a important criterion of the success of the Grid. The article brings out the criteria of evaluation of the function and performance of Grid resource and mechanism of optimization. The article uses a kind of resource pre-reserved mechanism, which is based on optimized-time-axes, it improves the acceptance of requests and the usage of effective resources.更多还原