MPLS-Kit: An MPLS Data Plane Toolkit

• Published in: 2022 IEEE 11th International Conference on Cloud Networking (CloudNet) pp. 49 - 54
• Main Authors: Vanerio, Juan, Schmid, Stefan, Schou, Morten Konggaard, Srba, Jiri
• Format: Conference Proceeding
• Language: English
• Published: IEEE 07.11.2022
• Subjects: Computational modeling; Data models; Multiprotocol label switching; Network topology; Topology; Virtual private networks; Web and internet services
• ISSN: 2771-5663
• DOI: 10.1109/CloudNet55617.2022.9978791

Networking research often requires a means to quickly generate different realistic networks for evaluating the practical relevance. This is especially the case for emerging fields related to the automated verification of network configurations ("what-if analysis") or to AI-driven network operations ("self-driving networks"). Unfortunately, the data of real world network deployments are often scarce. In particular, while the topologies of many real communication networks have been made available online, this data typically does not include the routers' forwarding tables, e.g., by Internet Service Providers (ISPs). This introduces a dilemma, as generating arbitrary forwarding rules for these topologies may not adequately mimic network behavior.We present MPLS-Kit, a tool for the automated generation of realistic MPLS data planes. In particular, the tool supports an efficient generation of MPLS data planes following widely-deployed industry-standard control protocols on top of arbitrary network topologies. Notably, MPLS-Kit supports the instantiation of MPLS Fast Reroute and VPN services. It further supports packet-level simulations providing a rich set of statistics about the simulated data plane which can be used for numerous applications, like congestion, latency, and resilience analysis. The generated data planes can be further exported in standard exchange formats and analyzed by formal verification tools.