A Service Path Selection and Adaptation Algorithm in Service-Oriented Network Virtualization Architecture

With the advances of networking and cloud computing technologies, many web services are often implemented as cloud services by different service providers, and some SPs compose those services to create an end-to-end network service. In this service-oriented network virtualization environment, two important but somehow contradictory technical challenges are 1) how to select an optimal service path in order to meet the QoS requirements by users, and 2) how to balance the loads to fully utilize the whole systems. In this paper, we present an adaptive service path selection algorithm in service-oriented network virtualization environment, which assures QoS and balances the load at the same time. The proposed algorithm selects appropriate component services from multiple candidate service instances running on virtual machines and creates an optimal service path satisfying the users' QoS requirements. Our algorithm also dynamically adapts to the optimal service path based on various changes. Since the proposed algorithm handles a variation of the multi-constrained path selection problem known as NP-complete, we formulate the problem using ant colony optimization algorithm. The experimental results show that our algorithm guarantees the maximal QoS to the users while balancing the load at the same time, and adapts to the optimal service path based on the situational changes.