Cloud Configuration Management With Constraint Solving

Cloud computing,based on the Internet,provides users with shared hardware and software resources and information according to their requirements.In cloud comput-ing pattern,cloud providers allow developers to put the applications in their cloud envi-ronments by providing a variety of functional cloud platforms.Developers leverage the infrastructure to deploy and configure applications for their personal requirements or enterprise business.However,deploying an application in a suitable cloud platform is not easy.Indeed,there exists tens of those cloud platforms,providing different supports and services as a wide range of resources at different level of functionality.Generally,it is up to developers to specify and exploit these characters to their best knowledge.Thus it is a high risk that results do not meet the expected requirements due to the lack of knowledge in specific domains.In order to help developers to select and configure cloud environments in a short time,we propose an approach based on software product line principles.With such an approach,we introduce a configuration mechanisms using feature models and a rea-soning mechanisms using constraint solving.Our work can be summarized as follows:?Leveraging the software product line principles,we separate the tasks of cloud providers and developers in cloud configuration management.Cloud providers fo-cus on describing the characteristics of cloud environments while developers con-centrate on requirements.?As part of the software product line principles,we leverage feature models for mod-eling cloud environments,which is suitable when considering commonalities and variabilities across various cloud platforms.To address limited expressiveness,we add attribute and cardinality extension to the features and constraints of original feature model.?We use Alloy modeling tool to reason on feature models.Constraint solving,as part of Alloy tool,is the core of reasoning process.We proposed two solutions in all,one for checking the validity of instances and the other for searching similar valid instances for invalid one.The latter case aims to give developers some useful feedback.?We design and implement SFEA system based on above techniques.We evaluate SFEA in expressiveness,efficiency and practicality.The results show that SFEA is capable of describing characteristics of different cloud platforms,it is efficient in reasoning and it helps developers in the deploying and configuring application process.