Research On Negotiation And Monitoring Of Service Level Agreement For Cloud Platform

Curently, cloud computing has provided a flexible, on-demand, dynamically scalable computing architecture for a variety of applications. However, the essence of services in a cloud environment is completely different from services in the traditional Internet, and user expectations for services are also different. Although Service Level Agreement(SLA) on the Web server has been very mature, but cloud service requires a different SLA management framework from Web services. In this context, SLA management for the cloud platform has a more important significance.SLA management for the cloud platform can be achieved by Broker, and it involves a lot of management functions, two important ones of which are SLA negotiation and SLA monitoring. Starting from the cloud service broker, which is a service agency that helps optimize the relations between cloud users and cloud service providers, this paper studies the architecture of the cloud service broker, and based on this, the game theory and the queuing theory are introduced to the study on SLA negotiation and SLA monitoring.This paper first proposes and studies the architecture of the cloud service broker,analyzes the role of cloud services and its key features, allowing cloud service agents to help cloud users find the most suitable cloud service providers through negotiation of SLA parameters. And the game theory is applied to cloud service broker, and the game process of SLA negotiation for the cloud platform is given, including strategy implementation for both sides of the game and bi-directional feedback of the game results. Then the game methods of SLA negotiation for cloud platform are proposed,including zero-sum game method between cloud users and cloud service agents and zero-sum game model between different cloud service brokers, and the specific strategies of Nash equilibrium and repeated game are studied separately. Furthermore,the queuing theory is applied to cloud service broker, the proposed queuing model uses the parallel type of multiple cloud service agents, and when the queuing rules for the cloud users are a waiting one, a losing one and a mixed one, the statistical research on the arrival of the cloud users and the service time are respectively performed, and the statistical rules are gained. The ultimate aim of this paper is to lay the foundation of SLA management for cloud platform.