基于约束求解的云平台配置管理技术研究

发布时间：2018-05-27 15:18

本文选题：云平台 + 软件生产线　；参考：《南京大学》2017年硕士论文

【摘要】：云计算是一种基于互联网的计算方式,通过这种方式,共享的软硬件资源和信息可以按需求提供给计算机各种终端和其他设备。在云计算中,云供应商允许开发者将写好的程序放在其提供的平台当中运行。应用开发者可以方便的部署各种应用来满足业务需求。然而,将应用部署到一个合适的云环境中并不是想象中的那样容易。面对众多云供应商,并且每个云供应商分别在不同层面提供功能和资源时,应用开发者往往很难做出合理的选择,通常会根据有限的专业知识进行判断和尝试。在尝试过程中,很有可能会因为在某个领域专业知识的匮乏而做出错误的判断。为了能在短时间内为应用开发者寻找一个合适的云平台来部署应用,并且该平台可以同时满足应用的功能性需求和非功能性需求,本文提出一个基于软件生产线技术和约束求解技术的解决方案。具体工作包括:●利用软件生产线技术,将云平台配置管理问题中应用开发者与云供应商的任务分解开来。其中,云供应商专注于描述云环境特点,应用开发者专注于描述应用需求。●使用特征模型对云平台进行建模。特征模型建模与平台无关,可以很好的应对云平台之间的异同性,并且特征模型自身具有良好的扩展性。在此基础之上,为了增强其表达能力,本文对原始特征模型的特征和约束增加了性质和基数两个方面的扩展。●使用Alloy建模工具,以约束求解的方式对模型实例进行有效性检测,提出两种检测方法。其中一种方法用于快速检测模型实例有效性,另一种方法用于当待检测实例无效时,通过搜索并反馈几组相似有效实例的方法来帮助应用开发者修改当前配置实例。●设计并实现了 SFEA系统,它将上面提到的众多技术集于一体。本文用实验证明SFEA系统有足够的表达能力去描述不同的云平台特点,可以在短时间内进行有效性检测,以及能够在应用开发者部署过程中带来实质性的便利。
[Abstract]:Cloud computing is an Internet-based computing method, through which shared software and hardware resources and information can be provided to computer terminals and other devices according to their needs. In cloud computing, cloud vendors allow developers to run written programs on the platform they provide. Application developers can easily deploy various applications to meet business needs. However, deploying applications to a suitable cloud environment is not as easy as you might expect. In the face of many cloud providers, and each cloud provider provides functions and resources at different levels, it is difficult for application developers to make reasonable choices, and usually judge and try on the basis of limited professional knowledge. In the process of trying, it is possible to make the wrong judgment because of the lack of expertise in a certain field. In order to find a suitable cloud platform for application developers to deploy applications in a short time, and the platform can meet both the functional and non-functional requirements of the application, This paper presents a solution based on software production line technology and constraint solving technology. The specific work includes: using software production line technology to decompose the tasks of application developers and cloud vendors in the cloud platform configuration management problem. The cloud provider focuses on describing the characteristics of the cloud environment, and the application developer focuses on describing the application requirements, and uses the feature model to model the cloud platform. The modeling of feature model is independent of platform, which can deal with the similarities and differences between cloud platforms, and the feature model has good expansibility. On this basis, in order to enhance its ability to express, this paper extends the features and constraints of the original feature model to two aspects: property and cardinality. Two methods are proposed to detect the validity of the model by solving the constraints. One is used to quickly detect the validity of model instances, the other is used to detect instances that are not valid, By searching and feedback several similar valid instances to help the application developer modify the current configuration instance, the SFEA system is designed and implemented, which integrates many of the technologies mentioned above. In this paper, it is proved by experiments that SFEA system has enough expressive ability to describe the characteristics of different cloud platforms, which can detect the validity of different cloud platforms in a short time, and can bring substantial convenience in the deployment process of application developers.
【学位授予单位】：南京大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.09

【相似文献】