TopoGen: A network topology generation architecture with application to automating simulations of software defined networks

Abstract

Simulation is an important tool to validate the performance impact of control decisions in Software Defined Networks (SDN). Yet, the manual modeling of complex topologies that may change often during a design process can be a tedious error-prone task. We present TopoGen, a general purpose architecture and tool for systematic translation and generation of network topologies. TopoGen can be used to generate network simulation models automatically by querying information available at diverse sources, notably SDN controllers. The DEVS modeling and simulation framework facilitates a systematic translation of structured knowledge about a network topology into a formal modular and hierarchical coupling of preexisting or new models of network entities (physical or logical). TopoGen can be flexibly extended with new parsers and generators to grow its scope of applicability. This allows to design arbitrary workflows of topology transformations. We tested TopoGen in a network engineering project for the ATLAS detector at CERN. 1 INTRODUCTION Operational computer networks are subjected to frequent reconfigurations in an effort to maintain their quality of service under uncertain conditions (produced by hardware, software or human failures). Meanwhile, the rapidly emerging Software Defined Networks (SDNs) technology offers an unprecedented capability to automatically and programmatically reconfigure large network topologies without the intervention of human operators. This new flexibility comes at the price of error proneness: the point of failure gets now shifted to the software that decides on network reconfiguration actions. The verification and validation of SDN-based design options becomes key to minimize the risk of deploying a faulty system. Network simulation has long been an efficient tool to gain confidence with networks at design time. When network simulation models are built based on real, changing topologies, each modification implies the need for updating the simulation model accordingly. The standard practice is to upgrade topology descriptions manually for a given modeling and simulation tool of choice. Such manual changes can get considerably time-consuming and error-prone, in particular for medium-to large-sized networks.

Formal Modelling of Software Defined Networking

“…An alternative is a much more abstract approach such as CloudSim which is used to model large SDN data centres [6] but this is an understanding of network performance before building the model, and hence is not suitable for the type of network performance and security modelling proposed here. A different approach is taken using TopoGen in modelling of SDN topologies [30] where the focus of the tool is in supporting large topologies and then using hybrid modeling, simulation and control of data networks based on a hybrid DEVS (Discrete Event System Specification) formalism where some packets are modelled explicitly and some as flows [8].…”

“…Furthermore, very few formal methods take a quantitative approach which is required if performance is to be measured. The hybrid simulation approach using DEVS and TopoGen [8,30] is the closest to our research and works with network specification languages that could be used to obtain the CARMA network description. We believe that our style of light-weight modelling style has a role to play, and the next phase of development is to embody our approach in software that conceals the details of the CARMA from a user without experience of formal methods to provide a software tool which enables specification of network topologies and traffic patterns in a simple format with graphical elements where appropriate.…”

“…Both of these approaches consider the full network stack, which can be very expensive in terms of initial setup, as well as computational resources to execute. Other simulation approaches allow for some abstraction [6,8,30] from these details, in order to reduce these overheads and various formal approaches have also been suggested [3, 4, 7, 27, 33-36, 40, 42], most of which have no quantitative features.…”