Dynamic Service Function Chaining by Resource Usage Learning in SDN/NFV Environment

Abstract

Recently, to reflect diverse service requirements increasing with the popularization of various Internet devices, network functions virtualization (NFV) is attracting attention as the core technology of the next-generation network. In keeping with the progress of NFV, service function chaining (SFC) for specific network service appeared, and SFC refers to a technique for the sequential abstraction of service functions. In connection with the existing study's method for dynamic service chaining that dynamically generates a service chain through reinforcement learning, considering a node, at which service functions operate for efficient service chaining in the NFV environment, and the usage of resources such as CPU, memory, and network usage, this paper investigated a method for more stable dynamic service function chaining by flexibly calculating a fixed weight for the $r$ value, which was derived from an experiment, according to the amount of remaining resources at the relevant node.