云计算环境下的Web服务管理技术研究

发布时间：2018-07-11 13:04

本文选题：Web服务 + 云计算　；参考：《国防科学技术大学》2014年硕士论文

【摘要】：随着资源服务化和服务异构化的发展,Web服务作为SOA(面向服务体系架构)的一种实现技术,当前,已经得到了广泛的应用。将Web服务运行在云计算平台之上,监测服务运行状态,通过资源按需供给可以使得服务动态调控资源。随着云计算平台下服务数量的快速增长,暴露了服务管理难、维护难等问题,Web服务管理成为需要解决的问题。本文首先构建了云计算环境下的Web服务管理模型,并围绕服务的配置管理技术和性能管理技术展开深入研究。主要工作包括:(1)云计算环境下的Web服务管理模型本文首先提出一种基于第三方的服务管理模型,用于管理云计算环境下运行的Web服务,从应用层、控制层、数据层和物理基础设施层对Web服务进行管理。从管理对象和管理操作两个方面进行了详细的描述。同时,考虑跨云平台服务管理的需要,设计了基于第三方跨云平台服务管理模型,该模型在基础服务管理模型之上,增加了配置管理协商模块,命令传输加密模块及权限跟踪模块,保障了跨平台管理器通信的安全,确保了服务管理器不会越权执行跨平台操作。(2)Web服务配置管理技术对UDDI注册中心的功能进行了扩展,从管理系统数据库获取服务相关的性能数据,能够实时展示服务运行的基本状态,从而使用户能够从注册中心获知更加准确完整的服务信息,有利于服务的选择。为了能更好地发现未注册的服务并对其实现全面的管理,使用DPI(深度报文解析)技术对网络数据包进行分析,通过关键字提取等方法识别服务,并将其加入到统一的服务信息库。为了能够快速调控服务和方便用户选择服务,使用DPI技术对网络报文数据进行分析,提取关键字,识别服务及调用关系。同时,扩展注册中心功能,提供服务关系注册功能,更好地发现服务关系。为了能够更好地提供给用户合适的服务,设计了一种基于社会关联和用户效能的服务选择算法,将服务与提供商、提供商与提供商之间的社会属性作为服务选择的参照依据之一,从用户的利益出发,根据用户对服务属性的模糊偏好,将服务的属性值量化,计算每个属性权值和综合值,依据综合值对服务排序,选择综合值最大的服务,改进了服务选择算法未考虑服务的社会属性和用户模糊选择的不足。(3)Web服务性能管理技术Web服务性能管理主要包括性能管理需求的配置功能,服务性能状态监测和服务资源调度,可以根据当前服务的运行状态动态申请或归还资源,以实现在云计算环境下的服务性能调控。包括被动调控和主动调控两种方法:被动调控首先设计Web组合服务的性能瓶颈识别算法,发现服务违反SLA的时候,通过分析组合服务中的每个子服务响应时间的变化率和所需资源的变化率,从而快速地定位需要调控性能的子服务及确定短缺的资源,向云运营商管理器提出增加对应服务虚拟机请求,为瓶颈服务提供所需资源;为了应对负载快速变化导致服务性能下降或资源长期浪费状况,提出了一种基于历史数据和服务状态转换进行负载预测的主动调控服务性能方法,将服务性能划分为五个状态,对运行服务进行性能测试并分析,为每一个服务的五个状态确定阈值,实时监测服务运行状态,当达到既定临界值时,主动向云运营商管理器提出增加虚拟机请求,以避免或缩短服务违反SLA的时间。(4)构建原型系统结合服务配置管理和性能管理技术研究,设计和实现了云计算环境下的Web服务管理系统,通过基于开源云环境系统Open Nebula搭建了私有云平台系统,将多个服务运行在云平台下,并实现配置管理和性能管理功能。
[Abstract]:With the development of resource service and service isomerization, Web service is a kind of implementation technology of SOA (Service Oriented Architecture). At present, it has been widely used. Web service is running on cloud computing platform, monitoring service running state and resource on-demand supply can make the service dynamic control resources. With cloud computing The rapid growth of service quantity under the platform exposes the problems of difficult service management and difficult maintenance. Web service management has become a problem to be solved. Firstly, this paper constructs a Web service management model under the cloud computing environment, and focuses on the service configuration management technology and performance management technology. The main work includes: (1) cloud computing ring The Web service management model in this context first proposes a service management model based on third parties to manage the Web service running in the cloud computing environment. It manages the Web service from the application layer, the control layer, the data layer and the physical infrastructure layer. The detailed description is made from two aspects of the management object and the management operation. Considering the need of service management across the cloud platform, a service management model based on third party cross cloud platform is designed. On the basis of the basic service management model, the model increases the configuration management negotiation module, the command transmission encryption module and the authority tracking module, which ensures the security of the cross platform manager's communication, and ensures that the service manager will not exceed the authority. (2) Web service configuration management technology extends the function of UDDI registration center, obtains service related performance data from the management system database, and can display the basic state of service running in real time, so that users can get more accurate and complete service information from the registration center, which is beneficial to the choice of service. In order to better discover unregistered services and achieve comprehensive management, DPI (deep message parsing) technology is used to analyze network data packets, identify services through keyword extraction, and add them to a unified service information base. In order to quickly control services and facilitate users to choose services, DPI technology is used. Analyze the data of network message, extract key words, identify service and call relation. At the same time, expand the function of registration center, provide service relation registration function, find the service relationship better. In order to provide better service to users, a service selection algorithm based on social association and user efficiency is designed. The social attributes between service providers and providers and providers and providers are one of the reference basis for service selection. From the interests of the users, according to the user's fuzzy preference of service attributes, the attribute values of the service are quantized, the weights and the comprehensive values of each attribute are calculated, and the services are sorted and the best services are selected according to the comprehensive value, and the improvement of the service is improved. The service selection algorithm does not consider the social attributes of service and the shortage of user fuzzy selection. (3) Web service performance management technology Web service performance management mainly includes the configuration function of performance management requirements, service performance status monitoring and service resource scheduling, which can dynamically apply or return resources according to the running state of current service. Now the service performance control in the cloud computing environment includes two methods: passive control and active control: passive control first designs the performance bottleneck recognition algorithm of Web composite service. When the service violates the SLA, it analyzes the rate of change of each sub service response time and the change rate of the required resources in the combination service, so as to speed up the change rate of the required resources. The location needs to regulate the performance of the sub service and determine the shortage of resources. The Xiang Yun Operator Manager proposes to increase the request of the corresponding service virtual machine and provide the required resources for the bottleneck service. In order to cope with the rapid change of the load and the long-term waste of resources, a transformation based on historical data and service state is proposed. The performance method of active control service for load forecasting is carried out to divide service performance into five states, test and analyze the performance of the service, determine the threshold for the five states of each service, monitor the running state of the service in real time. When the established critical value is reached, the active Xiang Yun Operator Manager proposes to increase the request of the virtual machine. To avoid or shorten the time of service violation of SLA. (4) build a prototype system combined with service configuration management and performance management technology, design and implement the Web service management system under the cloud computing environment, build a private cloud platform system based on the open source cloud environment system Open Nebula, run multiple services under the cloud platform, and implement the matching. Set management and performance management functions.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【参考文献】