Dynamic bandwidth allocation in SDN based next generation virtual networks: a deterministic network calculus approach

Abstract

Software-defined networking (SDN), recognized as next-generation paradigm, decouples the network control plane from the data forwarding plane for a logically centralized controller, allowing rapid networking technology innovations to serve great varieties of users' applications. SDN empowers the evolution of Internet with Open-Flow (OF) and taking advantages of Network Virtualization (NV) to provide efficient service slicing strategies. One of key research issues in both SDN and NV environments is a lack of resource scheduling mechanisms in the physical infrastructure. The resource scheduling mechanisms should be highly capable of ensuring network stability to add further benefits to SDN based next generation virtual networks. We propose a service discipline of dynamic bandwidth scheduling (DBS) within OF switches that dynamically allocates data rates to flows regarding QoS and flow dynamics. Furthermore, we formulate a coherent analysis framework of scheduling disciplines as a mathematical model based on the deterministic network calculus to provide a fast characterization of deterministic service guarantees in SDN. Finally, simulations validate derived theoretical bounds from the analysis framework and confirm that the DBS discipline guarantees QoS of all flows through dynamic bandwidth allocation and ensures an efficient allocation of system bandwidth.