Multi-cloud service provision based on decision tree and two-layer Restricted Monte Carlo Tree Search

Abstract

Compared with traditional service provision methods, cloud computing has many advantages, such as on-demand services and flexible pricing systems, attracting widespread attention from both academia and industry. However, the abundance of cloud services makes it difficult for users to choose suitable ones. Furthermore, the cloud is usually a dynamically changing execution environment, requiring users to constantly switch service carriers for the best experience. Therefore, to improve the efficiency of service interactions in a multi-cloud environment and cope with the high concurrency demands, this paper proposes a multi-agent and distributed broker-based cloud service provisioning model. Equipped with a decision tree-based user preference learning module, brokers in the proposed model can continuously adjust their resource introduction strategy through training on historical data. In the process of cloud service interaction, brokers select the best cost-effective service for users within a limited time through a two-layer Restricted Monte Carlo Tree Search (RMTCS) algorithm. This paper develops a multi-cloud service interconnection system based on the Libcloud framework. Shielding the different interfaces of different cloud service providers in cloud virtual machine and storage services provides the most suitable services for users and realizes the unified cloud federation service supply. The experiment results prove that the proposed model achieves better performance over the traditional methods and can better meet the service requirements of users.