Panoptisim is a tool for planning a partial SDN deployment for Panopticon. Given a topology, packet-level traffic traces, and various resource constraints, Panoptisim computes the locations to deploy SDN switches in the given network topology.

To install Panoptisim, Python 2.7 needs to be installed on the system. Then it can be installed by executing

python2 setup.py install

which will pull the necessary dependencies and install Panoptisim.

To get an overview about the supported parameters of Panoptisim, run

panoptisim.py --help

To use the included sample topology dataset, you need to specify

panoptisim.py --pickletopo

To start a new simulation, the keyword new needs to be supplied to Panoptisim. To get an overview of the parameters for running a new simulation, run

panoptisim.py new --help

The following example command should run out of the box:

./panoptisim.py --pickle new --seedmapping 1 --seednextswitch 1 --tmsf max-50 --epsf 1 --epp 10 --tm 2004 --switchstrategy RAND --portstrategy default --maxvlans 512 --maxft 100000 --toupgrade 100

It will begin a simulation based upon the supplied picked topology annotated based on the supplied parameters (seedmapping 1, seednextswitch 1, tmsf max-50, and tm 2004). It will iterate over 100 switches in random fashion, at each step, trying to satisfy SDNc ports subject to vlan flow table and traffic constraints.

Panoptisim consists of three main classes. A input parser for the command line, a class for holding an internal representation of a resource-annotated network topology and a simulator that determines the hybrid deployment. Additionally, there is support for parsing device-level configuration files (cited in the publication) to build the topology.

This file contains the argparse parser, documentation on each argparse option. It is the starting point for parsing the topology and the entering the simulator main loop.

This class implements network topology including an annotated NetworkX graph instance that represents directed link connectivity, link capacity, and switch metadata. The topology keeps state of the network link utilization as the simulation runs.

Responsible for parsing device level configuration files into Python data structures. Used by EnterpriseTopology to build the topology. Due to license issues, we are not able to provide the original device-level configuration data. Instead, we provide pickled EnterpriseTopology instances which we generated using this EnterpriseData class.

This is the class responsible for running the simulation.

======================

• Dan Levin dan@badpacket.in
• Fabian Schaffert fabian@badpacket.in
• Marco Canini marco@badpacket.in