地学数据网络服务组合模型的研究

发布时间：2018-03-23 18:03

本文选题：地学数据　切入点：服务组合　出处：《中国矿业大学(北京)》2017年博士论文

【摘要】：网络技术促进地球探测信息技术不断向着网络服务的方向发展,而云计算技术更加推动空间数据共享和交互操作向着应用服务的方向发展。随着地球探测过程中越来越复杂的数据采集、数据处理和数据解释等业务过程,网络服务技术中单个的原子服务提供的业务流程处理功能有限,无法满足用户的需求,需要通过各种分布式个体服务组合起来才能实现高质量和复杂的地学业务功能。为了确保地学数据应用系统运行的高效性,提前发现应用系统中潜在的错误以减少系统重新部署需要的成本等,满足复杂的业务功能需求和有效部署在云计算平台中,地学数据服务的建模分析显得尤为重要和日益迫切,地学数据服务组合的模型被提出作为地学信息服务的研究基础。基于空间数据的特征和网络服务技术基础上,研究首先构建地学数据服务模型,提出基于Petri网的地学数据服务组合模型,完成地学数据服务组合模型的正确性验证分析,然后完整描述地学数据服务组合网络应用系统的建模分析和系统实现,最后在地学数据服务组合模型的基础上,将地学数据服务组合模型部署到云计算平台中,开展云计算平台中地学数据服务组合的部署策略研究和优化研究。研究的主要成果包括如下五个部分:1.提出了基于Petri网的地学数据服务组合模型,并从可达性,死锁性,有界性和优化性方面验证模型的正确性。地学数据服务的描述是基于基本的服务实体的描述基础上的。地学数据服务组合的建模需要服务的描述和服务之间的关系描述,地学数据服务组合的建模过程是基于服务网络建模方法和四种基本的结构模式完成的,用于详细表述异步和并发的复杂地学数据服务组合过程。在将服务、服务组合应用服务网建模之后,服务组合模型的正确性的验证问题就转变成服务网的活性、有界性和死锁性等的验证。基于Petri网的地学数据服务组合模型及其正确性验证,可以在地学数据服务组合模型的建模阶段发现潜在的错误,避免地学数据服务组合模型在运行阶段错误执行,能够缩短运行阶段查找错误的时间,减少重新部署所需要的成本,增加业务流程的可实现度,以较低的成本达到整体最优。2.完整的描述了地学数据服务组合网络应用系统的建模阶段和实现阶段。三维地质模型的切割过程是一个典型的业务过程,研究中以地质模型切割的业务过程为例,完整的描述了地学数据服务组合网络应用系统的建模过程和实现过程。为了实现应用系统执行的高效性和准确性,结构的完整性是必须的。建模阶段集中在地学数据服务组合的建模和分析方面,实现阶段集中在地学数据服务组合网络应用系统的执行方面。在提出的服务网基础上,复杂的服务流程可应用服务组合网的方法建模。首先对地学数据服务组合建模并且验证分析其正确性,然后以三维地质模型切割为例,提出地学数据服务组合网络应用系统设计的通用框架,最后描述地学数据服务组合网络应用系统的实现过程。服务组合应用于地学数据网络应用系统的优势,包括成本低,效率高,易于应用,灵活性,可复用性和易于部署等。3.研究了云计算平台中地学数据服务组合的部署策略问题。地学数据服务组合模型及其网络应用系统在构建和实现之后,被部署到云计算平台中。云计算平台中地学数据服务组合网络应用系统高效运行的核心问题是如何选择最优的服务组合成复杂的服务,执行服务组合,以满足复杂的功能需求和实现高服务质量,被归结为优化问题。服务组合的部署策略是影响地学数据服务组合质量的一个很重要的因素,因此采用何种合适的部署策略是一个重要的问题。本研究中考虑的服务质量因素包括成本和响应时间,首先在应用系统和服务之间建立关联模型,应用有向非循环图来描述部署在云计算平台中的应用系统的复杂服务之间的关联关系,然后服务部署问题被映射为图的k分割优化问题,最后应用两阶段方法解决图的分割优化问题。一系列的实验验证了所提出的服务部署策略的可行性和有效性,所提出的服务部署策略明显优于改进的贪心算法,其中改进的贪心算法经常用于图的分割问题。4.构建云计算平台中网络拓扑结构优化的模型。在进一步探讨云计算平台中地学数据服务组合的优化研究之前,需要构建云计算平台中的网络拓扑结构优化模型。云计算平台中网络拓扑的优化问题描述为,云计算系统中存在成千上万的处于不同地理位置的服务器,如何将服务器有效组织是云计算系统高效稳定运行的关键问题之一,被归结为网络拓扑优化问题。考虑到云资源提供者和云用户,抽象出通用的云计算平台中网络拓扑结构,提出了以成本低和路径最短为目标函数的网络拓扑结构优化的模型,应用遗传算法和模式搜索算法的混合算法优化。遗传算法关注全局最优解而模式搜索更关注局部最优解。在应用混合算法的过程中,云计算平台中网络拓扑结构的鲁棒性得到验证。5.探讨云计算平台中地学数据服务组合的优化算法研究,在已有的理论方法研究基础上,统一服务质量QoS参数,解决从服务池中选择合适服务,提出服务组合限制,确定重要服务质量QoS,最终提出云计算平台中地学数据服务组合优化的算法。探讨云计算平台中地学数据服务组合的资源调度算法研究,重点分析服务质量QoS要求,识别最佳工作负载-资源对,为云工作负载调度合适资源。在资源供应方面,根据云用户的服务质量QoS要求为给定工作负载识别足够资源。在资源调度方面,根据资源供应选择的资源映射执行云用户工作负载。在已有的不同资源调度标准和参数下的资源调度算法研究基础上,提出云计算平台中地学数据资源调度的算法。探讨云计算平台中地学数据服务组合的负载均衡算法研究,重点分析云计算平台中的负载均衡定量指标如响应时间,可扩展性,资源利用率,通量,容错和性能等。在已有的云计算平台中负载均衡算法的研究基础上,提出云计算平台中地学数据服务组合的负载均衡算法。
[Abstract]:Network technology to promote the development of earth exploration information technology continuously toward the direction of network services, and cloud computing technology to promote the sharing of spatial data and the development of interactive operation toward the direction of the service. With the application of data acquisition in the process of earth exploration is more and more complex, data processing and data interpretation of business process, a single web service technology in atomic service the business process function is limited, unable to meet the needs of users, through a variety of distributed individual services together to achieve high quality and complex to learn business functions. To ensure that the learning efficiency of data application system, to discover their potential applications in the system to reduce the system error re deployment needs cost. To meet the functional requirements of complex and effective deployment in the cloud computing platform, modeling data service analysis is particularly As an important and urgent, models of geo data service composition is put forward as the basic research of geological information services. Based on spatial data characteristics and network service technology on the basis of research first builds geoscience data service model, put forward Petri network learning model based on the analysis of data service composition, verify the complete study the combined data service model, implementation analysis and system modeling and complete description of the geo data services combination network application system, finally combining data service model based on the geological data, the service composition model deployed in cloud computing platform, cloud computing platform to carry out geological data service composition deployment strategy research and optimization research. The main results of the research includes five parts as follows: 1. proposed Petri network data service model based on the combination of science, and from the reachability, deadlock and boundedness The correctness of verification and optimization of the model. To describe data services is based on the description of the basic service entity. The description of the relationship between modeling combined data service needs and description of service, the modeling process of geoscience data service composition is to complete the service network modeling method and four kinds of based on the basic structure model, for detailed asynchronous and concurrent complex geo data service composition process. In the service, after the service composition application service network modeling, verified the correctness of the service composition model into service network activity, verification of circles and deadlock. The Petri network study of data service composition model and its correctness is verified based on the modeling stage can combine data services model in the discovery of potential errors, avoid data service composition model in operation Phase error, can shorten the operation time of the phase errors, reduce re deployment costs, increase the realization of business process, with low cost to achieve the optimal overall modeling stage.2. complete description of the data service composition of network application system and implementation stage. The cutting process of 3D geological model is a typical business process, business process in order to study the geological model of cutting as an example, a complete description of the modeling process of geo data services combination of network application and implementation process of the system. In order to achieve high efficiency and accuracy of application system implementation, structural integrity is a must. The modeling stage of centralized modeling and analysis science data service composition in the implementation stage of centralized data network application system in the service composition execution. In the service network based on complex Service process modeling method of application service combination network. Firstly, the geo data service composition modeling and analysis to verify its correctness, and as an example to 3D model cutting, put forward general framework of geo data services combination network application system design, implementation process and finally describe geoscience data service composition of network application system. A combination in the geo data network application system advantages, including low cost, high efficiency, easy to use, flexibility, reusability and deployment strategy can be easily deployed on the.3. cloud computing platform in geo data service composition. Geoscience data service composition model and network application system in the construction and implementation. To be deployed in cloud computing platform. Cloud computing platform in the core issue for the efficient operation of data service composition of network application system is how to select the optimal service group The synthesis of complex services, implementation of the service composition, in order to meet the functional requirements of complex and high quality of service, is formulated as an optimization problem. The service portfolio deployment strategy is a very important factor to influence quality of data service composition, so the appropriate deployment strategy is an important issue. The quality factors considered in this study include the cost and response time, first built the relation model between the application systems and services, application of directed acyclic graph to describe the relationship between the deployment of complex service computing application system platform in the cloud between the service deployment problem is then mapped to the optimization problem of graph K segmentation. The application of two stage method to solve the optimization problem of segmentation graph. A series of experiments to test the proposed service deployment strategy is feasible and effective, the proposed service deployment strategy in Ming Dynasty Improved significantly better than the greedy algorithm, the improved greedy algorithm is frequently used.4. segmentation map to build cloud computing network topology optimization model of the platform. Before optimization research in cloud computing platform to further explore the geological data service composition, need to build cloud computing optimization model of network topology description optimization platform. The problem of cloud computing platform for network topology, cloud computing in different geographic locations there are tens of thousands of server systems, how to effectively organize the server is one of the key issues of cloud computing system efficient and stable operation, has been attributed to the network topology optimization problem. Considering the cloud resource providers and cloud users, abstracts the common Cloud Calculation of the topological structure of the network platform, put forward to the low cost and the shortest path for network topology optimization objective function model, using genetic algorithm Hybrid optimization algorithm and pattern search algorithm. The genetic algorithm on the optimal solution and the pattern search is more concerned about the local optimal solution. In the process of application of the hybrid algorithm, optimization algorithm for cloud computing robust network topology in the platform is verified on.5. cloud computing platform in geo data service composition, based on the theory the existing methods of research, unified QoS quality of service parameters, select the appropriate solution service from the pool, the service composition, determine the quality of service QoS, finally put forward the cloud computing platform in the data service composition optimization algorithm. The resource scheduling of cloud computing platform in geo data service composition, key analysis of quality of service requirements of QoS, the best working load on load identification resources, scheduling appropriate resources for cloud work. In the supply of resources, according to the cloud user service quality The amount of QoS required for a given work load identification of sufficient resources. In resource scheduling, resource mapping according to the resource supply chosen to execute the cloud user workloads. Based on the research of resource scheduling algorithms in different resource scheduling and parameters under the existing standard, the cloud computing platform in the data resource scheduling algorithm. Research on load balancing algorithm in the discussion of cloud computing platform, data service composition, focus on the analysis of cloud computing platform in the load balancing quantitative indicators such as response time, scalability, resource utilization, throughput, fault tolerance and performance. In the existing cloud computing platform based on load balancing algorithm on the proposed load balancing algorithm in cloud computing platform in the data service composition.

【学位授予单位】：中国矿业大学(北京)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P628

【相似文献】