【作者】 何强；

【导师】 金海；

【作者基本信息】 华中科技大学 ， 计算机系统结构， 2010， 博士

【摘要】 随着越来越多计算、存储和信息等资源的加入到面向服务环境中,组合服务质量保障的重要性愈发明显。在服务组合中,组合服务由多个组件服务组成,组合服务的质量取决于组件服务质量的综合。由于服务组合分为服务查询、服务选择和服务交付三个阶段,因此,如何在服务组合的各个阶段为组件服务提供质量保障以保障组合服务的质量是一个重要而又亟待解决的难题之一。在服务查询阶段,传统的中央服务器查询方法在面临大量的带服务质量限制的复杂服务查询请求时会遭遇性能瓶颈。结构化的点对点技术能够为服务查询提供良好的可扩展性,却无法保证数据的可用性。在基于Chord的带服务质量限制的分布式服务查询方法Chord4S中,服务提供商可以发布服务的质量说明,用户可以在查询中添加质量限制。Chord4S改进了Chord的数据分发方式和查询消息路由算法,提高了数据可用性并保证查询性能。实验表明Chord4S在动态的面向服务环境中的实现了高效、高可用的带质量限制的分布式服务查询。在服务选择阶段,如果选择了服务交付能力不足的服务提供商,有可能会导致组合服务的质量无法得到保障。为帮助用户预测服务提供商的服务交付能力,ServiceTrust基于服务提供商过往表现对其声誉进行评估,帮助用户挑选值得信任的服务提供商,并抵御面向服务环境中的恶意声誉操纵和服务质量滥用攻击。实验表明ServiceTrust提高了服务交付的成功率,并很好地抵御了恶意声誉操纵和服务质量滥用攻击。同样是在服务选择阶段,由于服务提供商提供服务质量的能力和意愿是动态可变的,用户需要与服务提供商就组件服务的质量进行协商,以确保其对组合服务质量的要求得到满足。CASS (Combinatorial Auction for Service Selection)通过组合拍卖,允许用户和多组服务提供商就组件服务的多维质量进行灵活而高效的协商。实验表明CASS帮助用户和服务提供商明确了服务交付过程中组件服务的质量,进而保障了组合服务的质量。在服务交付阶段,当组件服务出现异常时,需要对其进行校正以保障组合服务的质量。基于工作流模式对服务校正的范围进行限制,并通过比较校正方案所能带来的收益来确定校正方案。实验表明该方法提高了进行服务校正时组合服务质量保障的成功率。为了实现高效的服务查询、服务提供商评估、服务质量协商和服务校正,采用基于自治代理的服务保障框架,通过自治代理之间的交互来实现服务质量保障操作的部分或全部自动化。更多还原

【Abstract】 As more and more computation, storage and information resources are being added into the service-oriented environment, it is important to address the issue of QoS-aware service composition. In the process of service composition, a composite service is composed of a group of component services. The quality of the composite service depends on the aggregation of the quality of the component services. Since the process of service composition consists of three stages, i.e. service discovery, service selection and service delivery, it’s imperative and challenging to provide QoS guarantee for the component services at different stages of service composition in order to guarantee the quality of the composite service.First of all, at the stage of service discovery, traditional centralized service discovery faces the problem of performance bottleneck and single-point failure when it comes to massive QoS-aware service discovery. Structured P2P technology provides scalability to service discovery but cannot guarantee the data availability. Chord4S provides a distributed QoS-aware service discovery. In Chord4S, service providers can publish QoS descriptions. Service consumers can incorporate QoS constraints in service discovery. In order to enhance the data availability and guarantee the routing performance, Chord4S improves the data distribution method and routing protocol of Chord. Experimental results demonstrate that Chord4S provides an effective and efficient solution to distributed service discovery.Second of all, at the stage of service selection, if the selected service providers are incapable of delivering component services, the quality of the composite service will be violated. ServiceTrust help predict service providers’capability of service delivery by evaluating their reputation based on their historic service delivery performance. ServiceTrust can also resist the attacks of malicious reputation manipulation and QoS abuse. Experimental results demonstrate that ServiceTrust improve the success rate of service delivery and can well protect service consumers from the abovementioned attacks.Another issue that needs to be addressed at the stage of service selection is that service providers’capability and willingness to provide QoS is dynamic. Service consumers need to negotiate with service providers in order to guarantee their requirements for the quality of the composite services can be met. CASS (Combinatorial Auction for Service Selection) enables flexible and efficient combinatotiral-auction-based negotiation between service consumers and providers over multidimensional quality of the component services. Experimental results show that CASS can help the service consumers and providers determine the quality of the component services, which guarantees the quality of the composite services.Third of all, at the stage of service delivery, faulty component services must be adapted aiming at guaranteeing the quality of the composite service. A workflow-pattern-based adaptation method is proposed to confine the adaptation scope and help select worthwhile adaptation solution. Experimental results demonstrate that the proposed method can improve the success rate of quality guarantee for the composite service.Finally, a QoS guarantee framework based on autonomous agent is proposed to facilitate partially or completely automated service discovery, service reputation evaluation, QoS negotiation and service adaptation through interactions among autonomous agents.更多还原