By introducing programmability, automated verification, and innovative debugging tools, Software-Defined Networks (SDNs) are poised to meet the increasingly stringent dependability requirements of today's communication networks. However, the design of fault-tolerant SDNs remains an open challenge.
Renaissance [1] considers the design of dependable SDNs through the lenses of self-stabilization — a very strong notion of fault-tolerance. In particular, it is based on algorithms for an in-band and distributed control plane for SDNs, which tolerate a wide range of (concurrent) controller, link, and communication failures. These self-stabilizing algorithms ensure that after the occurrence of an arbitrary combination of failures, (i) every non-faulty SDN controller can eventually reach any switch in the network within a bounded communication delay (in the presence of a bounded number of concurrent failures) and (ii) every switch is managed by at least one non-faulty controller.
This repository features a prototype implementation of the Renaissance [1] algorithm, using the Floodlight SDN controller and Open vSwitch. An evaluation of performance via Mininet is provided as well.
- Move to the folder floodlight_global and build it via the ant command.
- Move to the folder floodlight_local and build it via the ant command.
- While in the floodlight_local folder, start the local controller by typing java -jar target/floodlight.jar
- Open a new terminal and move to the folder where the network topologies reside, namely mininetScripts. Start the desired topology using the command sudo python "filename".py
- A mininet prompt will show up once the network has been started. In the prompt, use the command xterm h1 to open a terminal on host 1.
- In the newly opened terminal on host 1, move to floodlight_global and start it by using the command java -jar target/floodlight.jar
- You are now able to see the network being discovered by the global controller. To run more than one controller, repeat steps 5-6 for other hosts.
This version of the implementation is by Ivan Tannerud and Anton Lundgren, and it builds upon earlier work done by Michael Tran. Emelie Ekenstedt provided assistance along the way. This code was developed as part of two master thesis projects at Chalmers University of Technology, where Elad Michael Schiller served as the primary supervisor and Stefan Schmid was the external examiner who was active in the evaluation of the prototype.
[1] Marco Canini, Iosif Salem, Liron Schiff, Elad Michael Schiller, Stefan Schmid `Renaissance: Self-Stabilizing Distributed SDN Control Plane,' International Conference on Distributed Computing Systems (ICDCS) 2018. An earlier version can be accessed via https://arxiv.org/abs/1712.07697