Fair virtual bandwidth allocation model in virtual data centers

Network virtualization opens a promising way to support diverse applications over a shared substrate by running multiple virtual networks. Current virtual data centers in cloud computing have flexible mechanisms to partition compute resources but provide little control over how tenants share the network. This paper proposes a utility-maximization model for bandwidth sharing between virtual networks in the same substrate network. The aim is to improve the fairness of virtual bandwidth allocation for multi tenants, especially for virtual links which share the same physical link and carry nonfriendly traffic. We first given a basic model and then proved the model can be split into many sub models which can run on each rack-switch port. In the sub model, every physical link is associated with a fairness index constraint in maximize utility calculating. The goal is to limit the differences among the bandwidth allocation of the virtual links which share the same physical link. Experimental results show that the virtual bandwidth allocation between tenants is more reasonable.