Revenue-driven dynamic resource allocation for network virtualization

Abstract

Network virtualization has emerged as a powerful and flexible platform for future network. A major challenge is virtual network embedding problem that deals with efficiently mapping of virtual nodes and virtual links onto substrate network resources. However, resource quantity requested by virtual network service is related to time and revenue, which is not concerned by existing researches, resulting in the inefficient utilization of substrate network resources and lower revenue. To address this problem, autonomic resource allocation model based on revenue and Revenue-Driven Dynamic Allocation (called RDDA) based on time interval classification (i.e., short-term, medium-term and long-term) are presented for first time. Simulation results show that RDDA improves the substrate network resource utilization rate and the revenue of Infrastructure Provider.