【作者】 李磊；

【导师】 黄涛；

【作者基本信息】 中国科学技术大学 ， 计算机软件与理论， 2008， 博士

【摘要】 Internet的发展与普及为软件技术带来了新的思路与挑战,继推动人与人间、人与应用间交互模式的革新后,正日益成为软件实体间互连互通的重要媒介。Internet技术的革新对软件工程的发展起到了有力的推动作用,而软件工程的发展带来了软件商业模式的变化。传统软件商业模式,即出售软件使用权(License)的模式已经难以满足未来的需求。以产品为中心向互联网客户为中心模式转变引发了下一代软件商业模式,即按需服务的模式。这种软件商业模式得到了工业界和学术界的高度重视。按需服务理念在工业界中的应用需要与之相配的计算范型的支撑。而传统计算范型面对的是相对静态、封闭的计算环境,难以应对Internet计算环境和业务敏捷性需求带来的新挑战。面向服务计算(Service Oriented Computing,SOC)正在成为新一代计算范型,SOC描绘了服务以开放、自主的方式运行在分布节点上,通过跨Internet的互联、互通、协同、联盟构造应用,系统能够根据环境变化和业务需求动态适应的新型应用图景。SOC利用服务作为基本构造单元,支持分布式应用的快速、低成本的组合式开发。众多良好特性使得SOC成为解决按需服务理想的关键技术。在用户需求的解决过程中,整个业务流程对用户应该是透明,用户不关心后台逻辑是如何组织、如何执行的,而仅要求自动高效的返回正确的解决方案。可以说业务流程的自动高效完成对实现按需服务的商业模式具有重要意义,是按需服务的商业模式能够在实际中应用的技术基础。然而,目前存在着很多关键技术问题制约着流程的自动完成,降低了解决方案的求解性能,使得用户需求无法得到自动高效的回应。而业务流程的自动高效对于应用逻辑提供商在商业领域的成功具有非常关键的作用,如果不能提供有效的保障机制,可能会造成服务不能满足预期需求,最终导致客户满意度下降。尤其在军事、金融、医疗等对性能要求较高的应用场景中,如果无法高效完成用户需求将会造成难以估计的损失。因此,研究自动高效的服务计算技术具有重要的学术和工业应用意义,本文针对面向服务计算中存在的若干关键技术问题进行了分析和研究。首先,在服务查找阶段。多QoS约束的组合服务选择问题是NP难解问题,其研究难点在于解空间过于庞大,无法在多项式时间内完成搜索。这导致了多QoS约束的组合服务选择无法在时间紧迫的应用中使用。本文提出了基于Web服务语义关联信息构建组合服务选择约束模型的方法。该约束模型能够有效的降低搜索空间,提高算法性能,并能够使得所选组合服务集有着良好的协作性。最后本文设计了高性能的启发式算法,能够高效的求得可行性组合服务集。此外,现有的单个服务发现方法的灵活性较差、匹配效率较低、实用性不强,且选择过程需要用户参与,无法做到自动执行。在实际应用中,用户迫切需要的是轻量级的、可以灵活有效的支持基于功能语义服务搜索的服务发现方法。本文提出了基于基因序列的服务匹配算法,以及主动式用户需求语义抽取技术,支持语义级匹配,有效保障了服务选择的精度。并提出了新式的服务注册选择的5层软件体系结构,为服务注册选择系统的设计提供了一种新的思路。其次,在客户端生成阶段。基于消息交换模式的客户端生成是目前的一个难点,研究灵活可扩展的消息交换模式自动实施框架是一个亟待解决的问题。本文基于形式化的消息交换模式描述方法能够严格刻画消息交换模式,消除现有描述方法所存在的潜在歧义性;基于自动机理论的消息交换模式处理机自动控制消息交换过程,为消息交换模式的自动实施提供了基础,并通过代码自动生成技术实现了消息交换模式处理程序的自动部署和加载。最后,在服务逻辑执行阶段。Web服务在继承XML优点的同时,也继承了XML低性能的缺点,这导致在对性能有较高要求的应用中,Web服务无法作为一个可以信赖的选择。SOAP消息处理的低效是导致Web服务性能低下的关键因素,而基于Java反射技术的数据绑定是Web服务的主要性能瓶颈。本文提出了一种新颖的Web服务性能优化策略。避免了传统的XML解析和运行时的Java反射过程,可在对消息的一遍扫描中完成SOAP处理,并且所有的预处理操作都在系统启动或服务热部署时完成,极大地提高了Web服务的性能。另外QoS对于Web服务在商业领域的成功具有非常关键的作用,如果不能提供有效的QoS保障机制,可能会造成服务不能满足预期需求,最终导致客户满意度下降。目前还缺乏有效的解决方案来为没有QoS能力的Web服务添加QoS保障,同时不改变原有服务的内部实现。如何在运行时为服务动态添加QoS保障能力是一个亟待解决的问题,它影响着整个流程的自动完成。本文提出了Web服务QoS动态保障技术,可以使原系统在不改变内部实现的情况下,向客户提供满足其QoS需求的Web服务。更多还原

【Abstract】 With the development of Internet, more and more software technologies are needed to research. Enterprise applications should respond rapidly to change, to integration and reorganization of the way to adapt to the new business mode. The development of the software engineering has played a strong role in promoting this. The traditional software business mode, namely, the sale of software License is difficult to meet future demand. Center for the products to the Internet customer-centric paradigm shift triggered a next-generation software business model, that is, on-demand service model. The software business model has been the industrial and the academia is highly valued. But the traditional computing paradigm is relatively static, closed computing environment, and can not match the new challenges.Service-oriented computing (Service Oriented Computing, SOC) is becoming the next generation of computing paradigm. SOC use of a service as basic structural units, distributed application support the rapid, low-cost modular development. Web services (Web Services) is the most promising technologies to realize SOC. Web services have lots of advantages, such as the use of XML language to describe service interfaces and service calls, using Internet standards such as HTTP communication protocol transmission. The above good features enable SOC the key core technologies to solve on-demand service. In an on demand business, process should be transparent to users, users do not care about the background of logic is how to organize, how to implement, and the only requirement is automatically get the correct and efficient solution. At present, there are many key technical problems constraining the process automatically, efficiently solved. Automatic and efficient business processes is very important for application logic provider. If not provide effective mechanism, may result in services that fail to meet anticipated demand, which eventually led to declining customer satisfaction. Especially in the military, financial, medical and other high performance requirements of the application scenes, if not efficient completion of user needs will cause incalculable losses. Therefore, the study of automatic and efficient service technology is very important to academic and industrial. In this paper, we do some research on the area of automatic and high performance service-oriented computing.First, in the service selection stage, Multi-QoS constrained Web services selection is a NP hard problem and its difficulty lies in the huge solution space. The NP hard problem can not be solved in polynomial time, which results in a combination of multi-service QoS constraints can not choose the time of the urgent application in use. In this paper, we propose a high performance approach for multi-QoS constrained Web Services composition. The correlations of all the candidate services are collected to construct a constrained model, which can reduce the search space efficiently. By using the constrained model, a heuristic algorithm is proposed to find the feasible solution with high performance and high precision. We performed experiments to evaluate the validity and efficiency of the model. The results show that the same level of success ratio, our approach can achieve a lower level of computational complexity. In addition, the existing single service discovery methods have lots of problems and could not be executed automatically. In this paper, based on the gene sequence of services matching algorithm, as well as active semantic extraction technology, we proposed a novel service selection approach to support semantic level match.Secondly, in the client generation stage, Message Exchange Pattern (MEP) is a challenging problem in the interaction of Web services. However, it is difficult to enact the MEPs automatically, which means the Web services infrastructure has to hard-code the support for various MEPs and it is also difficult to extend. In this paper, we propose an automatic enactment of MEPs for Web services. The enactment consists of two parts. The first part is the formal description method for MEPs, which is used to define MEP accurately with no ambiguities. The second part is the automatic enactment framework, which will automatically enact a new MEP through a MEP process machine that is generated according to the MEP definition. New MEPs would become easy to be engaged in Web services.Finally, in the service logic executed phase. SOAP, based on Extensible Markup Language (XML), inherits not only the advantages of XML, but its relatively poor performance. This makes SOAP a poor choice for many high-performance web services. In this paper, we propose a new approach to improve Web services performance. Focusing on avoiding traditional XML parsing and Java reflection at runtime, we create a service-specific SOAP processor to accelerate execution. Through our experiments in this paper, we observed that our approach obtained about a treble performance gain (maximum) by incorporating the SOAP processor into the Axis SOAP engine. In addition with the rapid development of Web Services, providing quality of service (QoS) becomes a key factor to achieve success in business. At present, a lot of enterprises systems do not provide QoS-Enabled Web Services. However, modifying the implementation of those systems to satisfy the requirements of consumers is exorbitant. To solve this problem, this paper proposed a dynamic QoS guarantee approach, which can provide a flexible mechanism for managing and deploying QoS handler, and makes legacy systems provide QoS On-Demand Web Service to consumers without changing the implementation.更多还原