Unified and standalone monitoring module for NFV/SDN infrastructures

In the ongoing process of telecommunications systems softwarization, a key role is played by Virtual Network Functions deployed in Cloud environments and interconnected through Software Defined Networks. The sepa­ration between data, control and service logic changes the way in which those systems are designed, operated, and capitalized, and also enables elasticity, flexibility and cost-efficiency in the whole ICT ecosystem. However, dedicated and reliable monitoring tools are required to take full advantage of software-based telecommunication systems. Such tools are expected to effectively combine metrics about various assets to provide a global view over the infrastructure, offload control plane from monitoring tasks to improve scalability, and be seamlessly inte­grated with the existing control plane through the interface. In this paper, we propose a monitoring module that meets the requirements imposed by the softwarization process. We present the reference architecture for the proposed solution, along with the fundamental assumptions we have based our work upon, and a Proof-of-Concept implementation. Our solution has been designed as a standalone Virtual Network Function, fully decoupled from the existing control plane. As a result, new features and degrees of freedom are available, especially in terms of adaptability to the heterogeneous softwarized infrastructure, with no modifications needed in the existing software or hardware components. The proposed modular architecture, supported by proper interfaces, can be integrated with a variety of tools to collects statistics from different assets, relieving the control plane from the burden of computational-intensive monitoring tasks. In order to carefully validate the design of the proposed unified and standalone monitoring module numerous scenarios addressing various aspects and potential impediments were considered, in a series of experiments run on our internal testbeds and on an auxiliary public cloud environment. The obtained results prove that our monitoring tool provides significant advantages with respect to the existing solutions which are integrated in the control plane. Therefore, it is able to cooperate with sophisticated traffic steering and cloud management mechanisms operating on the combined network and computing resources. Furthermore, we demonstrate that our solution is easily portable, for instance to a public cloud environment.