csepanda/qm

Quagga NS-3 Distrubuted Model configurable via yaml file

QM // Quick modeling tool

Building

Prerequisites

Dependencies

• libyaml-cpp
• ns-3-dce (dce-linux-1.10)

For distributed modeling NS-3 and DCE should be builded with --with-mpi / --enable-mpi options.

Toolchain

• [email protected] or higher
• [email protected] or higher

Required environment variables

Name	Description	Example value
DCE_HOME	Absolute path to DCE directory	home/csepanda/Work/DCE
DCE_ROOT	Absolute path to built DCE	$DCE_HOME/build
DCE_PATH	PATH to built DCE's executable	$DCE_ROOT/bin:$DCE_ROOT/bin_dce:$DCE_ROOT/sbin

Build

cmake .
make

Running

Prerequisites

Required environment variables

Name	Description	Example value
DCE_PATH	PATH to built DCE's executable	$DCE_ROOT/bin:$DCE_ROOT/bin_dce:$DCE_ROOT/sbin
LD_LIBRARY_PATH	PATH to built DCE's libraries	$DCE_ROOT/lib:$LD_LIBRARY_PATH

Run

./qm-runner <YAML_CONFIG>

Other tools

./tools/examine-files.bash

Show all last runned model's processes in nodes and exit status

./tools/clean-simulation-files.bash

Remove all files created during the simulation

qm's Issues

Add Docker images bundling

Model YAML Configuration: Application's arguments via scalar string

Currently application's arguments are passed via YAML sequence:

    - binary: ip
      node: !Ref node04
      arguments:
          - route
          - add
          - default
          - via
          - 10.0.2.1
          - dev
          - sim0
      startAt: 15s

It will be nice to have ability to specify arguments via scalar string:

    - binary: ip
      node: !Ref node04
      arguments: route add default via 10.0.2.1 dev sim0
      startAt: 15s

Add MPI options command line arguments configuration

Currently MPI can be configured only via simulation configuration

simulationConfiguration:
    networkStack: linux
    stopTime: 800s
    enableMpi: true
    systemIdMarkerStrategy: manual

It will be nice to have following command line arguments for MPI configuration:
MPI modeling enabling: --enable-mpi
SystemIdMarkerStrategy specifying: --system-id-marker-strategy=<strategy_name>

Traffic generator configuration

Refactor MPI Usage workarounds

Following code in ApplicationInstaller and LinuxStackIpConfiguration should refactored:

        if (ns3::MpiInterface::IsEnabled() &&
            ns3::MpiInterface::GetSystemId() != ns3Node->GetSystemId()) { // TODO refactor
            continue;
        }

Implement MCL algorithm for System Id Markering