VPN topology abstraction service using centralized core capacity sharing scheme

This paper deals with the problem of sharing VPN service provider's core topology and link state information with its VPNs. This is provided as a service using topology abstraction principles. The topology abstractions are generated considering factors such as VPN SLA requirements, fairness and maximizing core network utilization. The paper deals with algorithms applicable to generate topology abstraction for the VPNs in a centralized manner. In this context we define a problem called the VPN core capacity sharing problem. We study this problem using algorithms based on multi-commodity flow theory. We observe from simulation analysis over several topologies that applying the maximum multi-commodity flow based partitioning scheme improves the call performance and network utilization statistics compared to a previously proposed topology abstraction scheme based on maximum concurrent flow theory.