Inter-Datacenter Virtual Capacity Services: Reality and Mechanisms

The recent developments in network programmability and flexibility, shape the environment for the creation and offering of networking capacity services. Through such services one will be able to reserve on demand raw network capacity and for a specific duration, in the form of virtual links or of virtual networks that interconnect branch offices, datacenters or devices around the world. In this work, we initially present the capacity services landscape as it is formulated and discuss on the relations with the cloud computing service model. Next, we propose optimal Integer Linear Programming (ILP) based mechanisms for matching network demands to networking capacity services, in the same way users’ computing requirements can be matched to virtual computing instances. We show through simulations that the proposed mechanisms can optimize the bandwidth usage, while minimizing the cost associated with the use of the network resources. We also investigate the effects of malleable capacity requests and of dynamic pricing.