SLA-based trust model for secure cloud computing

Abstract

Cloud computing has changed the strategy used for providing distributed services to many business and government agents. Cloud computing delivers scalable and on-demand services to most users in different domains. However, this new technology has also created many challenges for service providers and customers, especially for those users who already own complicated legacy systems. This thesis discusses the challenges of, and proposes solutions to, the issues of dynamic pricing, management of service level agreements (SLA), performance measurement methods and trust management for cloud computing.In cloud computing, a dynamic pricing scheme is very important to allow cloud providers to estimate the price of cloud services. Moreover, the dynamic pricing scheme can be used by cloud providers to optimize the total cost of cloud data centres and correlate the price of the service with the revenue model of service. In the context of cloud computing, dynamic pricing methods from the perspective of cloud providers and cloud customers are missing from the existing literature. A dynamic pricing scheme for cloud computing must take into account all the requirements of building and operating cloud data centres. Furthermore, a cloud pricing scheme must consider issues of service level agreements with cloud customers.I propose a dynamic pricing methodology which provides adequate estimating methods for decision makers who want to calculate the benefits and assess the risks of using cloud technology. I analyse the results and evaluate the solutions produced by the proposed scheme. I conclude that my proposed scheme of dynamic pricing can be used to increase the total revenue of cloud service providers and help cloud customers to select cloud service providers with a good quality level of service.Regarding the concept of SLA, I provide an SLA definition in the context of cloud computing to achieve the aim of presenting a clearly structured SLA for cloud users and improving the means of establishing a trustworthy relationship between service provider and customer. In order to provide a reliable methodology for measuring the performance of cloud platforms, I develop performance metrics to measure and compare the scalability of the virtualization resources of cloud data centres. First, I discuss the need for a reliable method of comparing the performance of various cloud services currently being offered. Then, I develop a different type of metrics and propose a suitable methodology to measure the scalability using these metrics. I focus on virtualization resources such as CPU, storage disk, and network infrastructure.To solve the problem of evaluating the trustworthiness of cloud services, this thesis develops a model for each of the dimensions for Infrastructure as a Service (IaaS) using fuzzy-set theory. I use the Takagi-Sugeno fuzzy-inference approach to develop an overall measure of trust value for the cloud providers. It is not easy to evaluate the cloud metrics for all types of cloud services. So, in this thesis, I use Infrastructure as a Service (IaaS) as a main example when I collect the data and apply the fuzzy model to evaluate trust in terms of cloud computing. Tests and results are presented to evaluate the effectiveness and robustness of the proposed model.