High Satisfaction and Fair Allocation of Resources in Software-Defined Data Center Networks

This paper studies how to fairly and efficiently allocate the two limited resources in software-defined data center networks (DCNs), namely the flow table and the control channel bandwidth. The problem considers routing path selection together with allocation of flow table entries and control channel bandwidths. The objective is to maximize the satisfaction ratio for flow groups in the network in terms of the two aforementioned resources, with the routing path optimized and different fairness constraints enforced. Our approach is to aggregate individual flows into flow groups and then find the optimal routing paths and the corresponding resource allocation vectors for each flow group. We study different fairness models in this work and also include a mechanism to relax the fairness constraint, which produces a range of solutions that permits a trade-off between total demand satisfaction and fairness.