Load-aware and congestion-free state management in network function virtualization
Jinwei Liu, Haiying Shen, Hongxin Hu · 2017 International Conference on Computing, Networking and Communications (ICNC) · 2017
Software-defined networks are constantly evolving due to the updates such as network function (NF) state updates, VM migrations. Network functions virtualization (NFV) with software-defined networking (SDN) has the capability of accurately monitoring and manipulating network traffic, and reducing operating cost. However, it cannot effectively handle the congestion and satisfy service level agreements (SLAs) on NF performance (e.g., throughput) while minimizing the operating cost in the scenarios of requirements for packet processing to be redistributed across a collection of NF instances simultaneously. Although OpenNF, a control plane architecture can allow quick, safe, and fine-grained reallocation of flows across NF instances, it neglects the congestion existing in practical scenarios for scheduling the updates, which can result in SLA violations. Also, it does not consider the load of links to schedule the updates for minimizing the operating cost. To address this problem, we adequately consider the congestion caused by the competition for limited resource (e.g., bandwidth) and utilize the load information to propose a load-aware and congestion-free state management (LCSM) strategy. LCSM can provide congestion-free scheduling of updates and minimize the operating cost. Extensive simulation results show the advantages of our proposed LCSM.