Automated Design and Dimensioning of Virtual Networks

Automated Design and Dimensioning of Virtual Networks

Siavash Khorsandi, Jim Yan


Future broadband networks will be dealing with communities of users with vast differences in their traffic management access, security, and quality of service requirements, Differentiated treatment of various communities of users results in several logical overlay subnetworks on top of a common physical infrastructure. 'This paper deals with configuration and dimensioning of these subnetworks on a dynamical basis in order to adapt the resource allocation of the VNs based on their traffic profile and service quality requirements. The objective is to maximize resource efficiency and hence network revenue for a given available physical resources. 'The parameters to be optimized are capacity allocations of logical links and traffic distribution factors that indicate how the total traffic between an end-node pair is distributed through available parallel routes, A general scaleable algorithm is proposed which is capable of dealing with complex constraints and traffic models. The resulting adaptive virtual network framework provides an efficient mechanism for supporting quality of service (QoS) in a high-speed multi-service environment as well as quick deployment of virtual private networks (VPNs). Although the initial focus of this work has been on ATM networks, due to the generic nature of the underlying subjects (e.g. resource allocation, differentiated service, etc.), it can also be a basis for QoS support in IP networks.


Virtual networks, flow optimization, capacity assignment, logical subnetworks