Profiling Service Function Chaining Behavior for NFV Orchestration

Abstract

The concepts of Software Defined Networks (SDN) and Network Function Virtualization (NFV) have promoted network chaining, or Service Function Chain (SFC), quickly and simply. In dynamic infrastructure scenarios, the management of SFC introduces challenges both for the connection of these elements and for understanding their behavior when automatic elasticity is required. Currently, most existing approaches have addressed this challenge with probabilistic heuristics or AI-based techniques, considering only static management. This paper presents an approach for profiling SFC that can be used for predictive NFV orchestration purposes. We conducted a performance evaluation study based on testbed experimentation and queueing modeling. Our results confirm that an analytical model can be used for managing SFC orchestration, not only as a validation technique for NFV, but also as a heuristic for predictive resource allocation in production environments.