Use of Parallel and Distributed Processing in the Control Plan of Software Defined Networks to Increase Energy Efficiency in Data Center Networks
Software Defined Networks, Energy Efficiency, Low Power, Parallel Processing, Distributed Processing, Data Centers
The main feature of the software-defined networks is the separation of the role of decision-making, known as control-plane, and the role of routing of the packages, known as data-plane. This separation allowed the introduction of the concept of network programmability, with which new applications could be implemented to interact directly with the operation of the networks. Today, these applications enable data center environments to match demand elastically, enabling cloud computing services. In this scenario, the datacenters became the prominent service providers, and one of its main costs is the consumption of energy in the infrastructure of servers and network equipment. Many papers indicate that software-defined network in datacenter networks allows for better energy efficiency, especially in the data plane. In this work, we present a strategy to – using parallel processing and distributed processing with lower operating frequency on the processing elements – reduce the energy consumption of the controllers on a software-defined network. The implementation of parallel and distributed versions of an SDN controller offers a fault-tolerant energy-aware solution to the presented problem.