Announcing Mininet 2.3.0 !

We are pleased to announce Mininet 2.3.0 !

This release provides a number of bug fixes as well as new features, including support for Python 3 (as well Python 2) and installation from source on Ubuntu 20.04 LTS as well as Debian 10+.

Please check out additional details on the Mininet 2.3.0 release in the release notes on docs.mininet.org.

How to Get Started

  • The easiest way to get started with Mininet is to download a VM image. See Download for details.

  • The latest Mininet source code is available on GitHub and also via code.mininet.org.

Mininet is an open source project, and we are grateful to our many code contributors, bug reporters, and active users who continue to help make Mininet a useful system with a friendly and helpful community.

We look forward to seeing what you will do to transform and improve the networking world.

Have fun with Mininet!

Mininet Wins ACM SIGCOMM Test of Time Paper Award

We are pleased to announce that our original HotNets paper, A Network in a Laptop: Rapid Prototyping for Software-Defined Networks, has won the ACM SIGCOMM Test of Time Paper Award.

Congratulations to the authors, Bob Lantz, Brandon Heller, and Nick McKeown, and congratulations and thanks to the many researchers, teachers, students, users, collaborators, and project contributors who have used and continue to use Mininet to help usher in the SDN revolution and improve the practice of computer networking!

Congrats to Mininet SIGCOMM Hackathon Winners!

Congratulations to Alexander Jung, Michał Król and Torsten Runge, who prototyped a Mininet node class for launching unikernels - small application-specific kernel images, basically an application linked with a library OS, ClickOS being one example - and ended up placing in the SIGCOMM 2018 Hackathon!

http://conferences.sigcomm.org/sigcomm/2018/hackathon.html

In their case they used Rumprun as the unikernel system and qemu/kvm to launch it:

They are planning on continuing to work on it, so I hope we can get something simple and useful running and perhaps added to Mininet!

Mininet Wins ACM SIGCOMM SOSR Software Systems Award !

We are pleased to announce that Mininet has won the ACM SIGCOMM SOSR Software Systems Award!

“The ACM SIGCOMM SOSR Software Systems Award is given to an individual or an institution to recognize the development of a software system that has had a significant impact on SDN research, implementations, and tools. The impact may be reflected in the widespread adoption of the system, or of its underlying concepts, by the wider SDN community, either in research projects, in the open-source community, or commercially.”

The award also includes an invited talk, and the talk slides may be found here.

The SOSR 2017 program and proceedings may be found here.

Mininet’s ongoing success and impact would not be possible without the support and contributions from the Mininet community, so we’d like to take this opportunity to thank everyone and to say that we look forward to working together to continue to make Mininet a better and more useful system!

Mininet VM Security Advisory

Unfortunately, installing the Ubuntu SSH package into a VM image causes default SSH keys to be generated. In the case of the Mininet VM, this means that the VM images for Mininet 2.2.1 and earlier included the same set of default SSH keys (which could leave you open to a MITM attack.) This probably isn’t surprising to many of you, but we’re changing it in the Mininet VM images to avoid this issue.

As of Mininet 2.2.2, we no longer ship default SSH keys in the provided Mininet VM; we remove them before upload, and new SSH keys should be generated the first time you boot a Mininet 2.2.2 VM image.

If you are logging into an older Mininet VM over the internet or another non-secure network, we highly recommend regenerating your SSH keys if you haven’t done so already:

1
2
3
sudo rm -f /etc/ssh/*key*
sudo /usr/sbin/dpkg-reconfigure openssh-server
sudo service ssh restart

Alternately, you are welcome to download a new Mininet 2.2.2 VM image.

If you’re distributing a VM image for any purpose, we recommend removing any SSH keys before uploading it!

Additional information on Mininet 2.2.2 may be found at http://mininet.org/blog/2017/03/17/announcing-mininet-2-2-2/

Announcing Mininet 2.2.2

Mininet 2.2.2 is primarily a performance improvement and bug fix release.

Security Advisory!

  • The Mininet VM images no longer ship with identical ssh keys. We recommend that you destroy any default ssh keys in existing Mininet VMs if you haven’t already:
1
2
3
sudo rm -f /etc/ssh/*key*
sudo /usr/sbin/dpkg-reconfigure openssh-server
sudo service ssh restart

Improvements and Bug Fixes

  • We now run public CI builds via Travis
  • Several common test failures have been addressed
  • We use our own forked versions of oflops and the Stanford Reference Switch/Controller
  • install.sh now has basic support for RHEL
  • mnexec now works on systemd systems
  • The default OpenFlow port is now the official port of 6653 rather than the classic port of 6633, which was nastily co-opted by Cisco.
  • Several issues with the NAT element have been addressed

Bugs and Gotchas

  • Ubuntu 16.04 LTS (Xenial) kernel bug: the problem resulting in unregister_netdevice: waiting for lo to become free. Usage count = 1 in the console or kernel log has returned. We recommend using 14.04 or an older kernel until this is fixed.
  • Ryu installation with install.sh -y is still broken
  • The --tests option is slightly different from mainline

Release notes and Downloads

Full release notes for this release and previous releases may be found in the Release Notes on docs.mininet.org.

The easiest way to get started with Mininet is to download a VM image.

Download, installation, and upgrade instructions may be found at http://mininet.org/download .

The latest Mininet source code is available on github and also via code.mininet.org.

Acknowledgments

Thanks to all of the contributors of fixes and enhancements for this release, including Olivier Tilmans, Jono Hart, Tomasz Buchert, Rahman Pujianto, Roan Huang, M S Vishwanath Bhat, Brian O’Connor, and others.

Automated Testing Using Travis CI

Mininet is now using Travis CI for automated testing!

Whenever code is pushed to Mininet, Travis CI starts a test job that installs Mininet, runs a simple sanity test, and the “quick” tests for HiFi, the walkthrough, and the example code.

The full results are available on Travis CI’s Mininet page, and the status of the most recent master build is part of the README on Github:

Travis CI also verifies all new pull requests and displays their build status in the pull request discussion.

Mininet will continue to be tested in several ways on private servers, but our hope is that having a public-facing CI will give members of the community a better sense of how Mininet is tested and the quality of the codebase. Furthermore, we hope that by automatically testing pull requests, we will reduce the likelihood of merging broken code and reduce the testing burden on maintainers.

Announcing Mininet 2.2.1

Mininet 2.2.1 is primarily a performance improvement and bug fix release.

  • Batch startup has been implemented for Open vSwitch, improving startup performance.

  • OVS patch links have been implemented via OVSLink and --link ovs

    Warning! These links have serious limitations compared to virtual Ethernet pairs: they are not attached to real Linux interfaces so you cannot use tcpdump or wireshark with them; they also cannot be used in long chains - we don’t recommend more than 64 OVSLinks, for example --linear,64. However, they can offer significantly better performance than veth pairs, for certain configurations.

Also:

  • You can now easily install Mininet on a Raspberry Pi (simply by using install.sh -fnv or install.sh -a) ;-)

  • mn now accepts multiple --controller arguments

  • --controller remote and RemoteController now accept ip arguments of the form <IP address>:<port> (in addition to ip=<IP>,port=<port>)

Additional information for this release and previous releases may be found in the Release Notes on docs.mininet.org.

The easiest way to get started with Mininet is to download a VM image.

Download, installation, and upgrade instructions may be found at http://mininet.org/download .

The latest Mininet source code is available on github and also via code.mininet.org.

Thanks to all of the Mininet contributors! A number of interesting features and improvements are already in the works for Mininet 2.3.

Mininet on Raspberry Pi

Mininet is now easy to install on the Raspberry Pi running Debian/Raspbian 7 Wheezy!

Mininet’s install.sh has been updated so it should simply work:

install.sh -fnv  ## basic install
install.sh -a    ## the works

Here is an example of Mininet running on a Raspberry Pi 2 model B:

A Raspberry Pi can be used any time you need a cheap network-in-a-box, for example for courses, laboratories, demonstrations, or simple experiments.

We look forward to hearing (on mininet-discuss and elsewhere) about interesting, fun and clever uses of Mininet running on the Pi !!

Announcing Mininet 2.2.0 !

We are pleased to announce Mininet 2.2.0! This release provides a number of bug fixes as well as several new features, including:

  • Improved OpenFlow 1.3 support

    • mn --switch ovs,protocols=OpenFlow13 starts OVS in 1.3 mode
    • install.sh -w installs 1.3-compatible Wireshark dissector using Loxigen
      • install.sh -y installs the Ryu 1.3-compatible controller
  • A new nodelib.py node library, and new Node types including LinuxBridge, OVSBridge, LinuxRouter (see examples/) and NAT

  • Easier LAN/internet connectivity from Mininet hosts

    • mn --nat will semi-automatically connect a Mininet network to your LAN using NAT, solving the “why can’t I ping Google?” problem.

      Warning: By default this will reroute local traffic originating at your Mininet server or VM and destined for Mininet’s IP subnet (10.0.0.0/8 by default) to the Mininet network, which can break connectivity if you are using addresses in the same range in your LAN. You can change this range using the --ipbase option.

  • An improved MiniEdit GUI (examples/miniedit.py) - thanks to Gregory Gee

  • Support for multiple --custom arguments to mn

    mn --custom mytopo.py,myswitch.py --topo mytopo --switch myswitch

  • Experimental cluster support - consult the documentation for details - as well as examples/cluster.py and an experimental --cluster option for topologies built with the default Host and OVSSwitch classes:

    mn --cluster localhost,server1,server2,server3 --topo tree,4,4

Note that examples contain experimental features which might “graduate” into mainline Mininet in the future, but they should not be considered a stable part of the Mininet API!

How to get it

The easiest way to get started with Mininet is to download a VM image. See http://mininet.org/download for details.

The latest Mininet source code is available on github and also via code.mininet.org.

Additional information on the Mininet 2.2.0 release may be found in the release notes and on docs.mininet.org.

Mininet is an open source project, and we are grateful to our many code contributors, bug reporters, and active users who continue to help make Mininet a useful system with a friendly and helpful community!

Have fun with Mininet!

Mininet 2.2.0 Core Team:
Bob Lantz, Brian O’Connor, Cody Burkard
Open Networking Laboratory

Minininet Tutorial at SIGCOMM

Materials from our tutorial at SIGCOMM 2014, “Teaching Computer Networking with Mininet”, are available on teaching.mininet.org. The original announcement follows.


Call for Participation:

Mininet Tutorial - Teaching Computer Networking with Mininet

http://conferences.sigcomm.org/sigcomm/2014/tutorial-mininet.php

August 18th, Chicago, USA (in conjunction with SIGCOMM)


Important Dates

  • Early Registration Deadline: July 11th, 11:59pm CDT
  • Late Registration: July 12th - August 18th
  • Tutorial Date: August 18th 8:30am - 12:00pm

Objective

Sharing our experiences using the Mininet network emulation platform to teach computer networking in a variety of contexts ranging from small on-campus courses (Stanford, Georgia Tech, MIT) to massive online courses. We believe that the Mininet platform can make teaching and learning computer networking more fun and effective by facilitating experiential learning.


Program

  • Introduce Mininet to the broader network community, including its capabilities and limitations and some example use cases.
  • Familiarize attendees with Mininet through hands-on exercises.
  • Share our experiences using Mininet since 2011 for on-campus and massive online courses at Stanford, Georgia Tech and MIT.
  • Disseminate assignments that we used at our universities and help to develop and expand an online repository of Mininet courseware.

Speakers and Panelists

  • Bob Lantz - ON.Lab
  • Te-Yuan (TY) Huang - Stanford
  • Vimalkumar Jeyakumar - Stanford
  • Brian O’Connor - ON.Lab
  • Nick Feamster - Georgia Tech
  • Keith Winstein - MIT
  • Anirudh Sivaraman - MIT

Announcing Mininet 2.1.0 !

We are pleased to announce the immediate availability of Mininet 2.1.0 on mininet.org !

Mininet is a network emulation platform that is intended to be the quickest, easiest, and most enjoyable way to create virtual networks for research, education, and system development. By providing an instant virtual network on your laptop that can run real application, switch, and controller code, Mininet also makes it convenient to get started with SDN and OpenFlow without any additional hardware.

Mininet 2.1.0 provides a number of bug fixes as well as new features, including:

  • Convenient access to Mininet() as a dict of nodes
  • X11 tunneling (wireshark in Mininet hosts, finally!)
  • Accurate reflection of the Mininet() object in the CLI
  • Automatically detecting and adjusting resource limits
  • Automatic cleanup on failure of the mn command
  • Support for installing the OpenFlow 1.3 versions of the reference user switch and NOX from CPqD and Ericsson
  • Preliminary support for running OVS in user space mode
  • Preliminary support (IVSSwitch()) for the Indigo Virtual Switch
  • Preliminary support for Fedora in install.sh
  • A script (util/m) for easily running new shells or commands in nodes
  • The ability to import modules from mininet.examples

We have provided several new examples (which can easily be imported to provide useful functionality) including:

  • Modeling separate control and data networks: mininet.examples.controlnet
  • Connecting Mininet hosts to the internet (or a LAN) using NAT: mininet.examples.nat
  • Creating per-host custom directories using bind mounts: mininet.examples.bind

Note that examples contain experimental features which might “graduate” into mainline Mininet in the future, but they should not be considered a stable part of the Mininet API!

Additional information may be found in the Release Notes.

Mininet is an open source project, and we are grateful to our many code contributors, bug reporters, and active users who continue to help make Mininet a useful system with a friendly and helpful community!

Have fun with Mininet!

The Mininet 2.1.0 Team
Bob Lantz and Brian O’Connor
Open Networking Laboratory

Nick Feamster's SDN Course

I just want to put in a recommendation for Nick Feamster’s excellent Software Defined Networking course. It’s free on Coursera, and not only does he do a great job of clearly explaining SDN, network virtualization, and related topics, he has also created some excellent videos explaining the usage, design, and rationale for the Mininet system, which is also being used for programming assignments and labs in the course.

There are also interesting interviews with notable people like Martin Casado, David Clark, and Teemu Koponen. It’s quite excellent, very approachable for anyone who has a basic networking background, and highly recommended:

https://www.coursera.org/course/sdn

The course has already started, but as of today (Thursday July 11th) you can still sign up for it.

-Bob

Automating Controller Startup

One of Mininet’s key features is that it makes it very easy to create a complete virtual network including hosts, switches, links and OpenFlow controllers. By default, Mininet runs Open vSwitch in OpenFlow mode, which requires an OpenFlow controller.

As you probably know, Mininet comes with built-in Controller() classes to support several controllers, including the OpenFlow reference controller (controller), Open vSwitch’s ovs-controller, and the now-deprecated NOX Classic.

You can easily choose which one you want when you invoke the mn command:

sudo mn --controller ref
sudo mn --controller ovsc
sudo mn --controller NOX,pyswitch

Each of the above examples uses a controller which turns your OVS switches into Ethernet bridges (aka learning switches.)

Note that ovsc is easy to install but only supports 16 switches. You can install the reference controller (patched to support many switches) using install.sh -f. (As of this writing, you can also install NOX Classic using install.sh -x.)

The manual way: using remote controllers

But what do you do if you want another controller like POX? Well, one way is to start up POX (or Beacon, Floodlight, etc.) in another terminal window and then use --controller remote:

sudo mn --controller remote,ip=127.0.0.1

If your controller is running locally, you can just use

sudo mn --controller remote

You might be asking: is there a way for Mininet to automatically start up my controller so that I don’t have to do it manually? The answer is, of course, yes!

The automatic way #1: using the command line

The latest Mininet master branch allows you to change the command and arguments for --controller ref to invoke another controller as desired. For example, here’s how you could invoke POX:

sudo mn --controller ref,command='python pox.py',cdir=$HOME/pox,cargs='openflow.of_01 --port=%s forwarding.l2_learning'

Note the following options:

cdir: the directory to change to before invoking the controller

command: the command to start your controller

cargs: the controller’s arguments (including %s for wherever the controller’s listening port should go)

The super automatic way #2: creating a custom Controller() subclass

This is much easier than it sounds, and it has worked in Mininet since version 1.0!

Here is a custom class for POX’s l2_learning:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
from mininet.node import Controller
from os import environ

POXDIR = environ[ 'HOME' ] + '/pox'

class POX( Controller ):
    def __init__( self, name, cdir=POXDIR,
                  command='python pox.py',
                  cargs=( 'openflow.of_01 --port=%s '
                          'forwarding.l2_learning' ),
                  **kwargs ):
        Controller.__init__( self, name, cdir=cdir,
                             command=command,
                             cargs=cargs, **kwargs )

controllers={ 'pox': POX }

With this class saved into pox.py, you can now do the following:

sudo mn --custom pox.py --controller pox

Additionally you can use the custom class in your Mininet scripts:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
#!/usr/bin/python

from mininet.log import setLogLevel
from mininet.net import Mininet
from mininet.topolib import TreeTopo
from mininet.cli import CLI

# We assume you are in the same directory as pox.py
# or that it is loadable via PYTHONPATH
from pox import POX

setLogLevel( 'info' )

net = Mininet( topo=TreeTopo( depth=2, fanout=2 ),
               controller=POX )

net.start()
CLI( net )
net.stop()

Note that in this custom class we’ve simply changed __init__, but in a more complicated example you might wish to change the start() and stop() methods as well.

Announcing Mininet 2.0.0

We are pleased to announce the availability of Mininet 2.0.0!

In the last few years, the Mininet community has grown from a few users at Stanford to an active mailing list with over 500 members in academia, startups, and industry.

Version 2 (“HiFi”) is a major upgrade that expands Mininet’s scope from functional testing (“does my network control plane work?”) to performance testing (“how well does my custom congestion control perform with 10 Mb/s links?”). New APIs now allow you to create links with specified bandwidths, latencies and loss rates, and to limit the CPU usage of virtual hosts. Using these new features, 37 students in Stanford University’s Advanced Topics in Networking course (CS244) replicated 16 published research experiments; you’re encouraged to browse their stories at The Reproducing Network Research Blog.

Other new features of Mininet 2.0.0 include:

  • An all-new website, http://mininet.github.com, with online doxypy documentation for all classes.
  • More flexible mn command line options, including the ability to specify NOX (and, in the future, POX) modules from the mn command line, and the ability to set link and CPU bandwidth limits
  • Dynamic link reconnection in OVS (supports migration and mobility emulation)
  • First-class abstractions/classes for link (Link) and interface (Intf) objects
  • Interface/link bandwidth limiting support using tc, via TCIntf and TCLink classes
  • CPU bandwidth limiting supporting both CFS bandwidth limiting and POSIX real-time scheduling, via the CPULimitedHost class
  • Easier task execution on virtual hosts via the Host.Popen() interface, implemented in mnexec using the setns system call
  • Easier and faster native installations that can use the latest Open vSwitch packages.

Note that the Topo object has changed somewhat to enable more flexible topology params, and as a result, your old code may need a few changes.

Download and installation instructions are available at http://mininet.github.com/download and Mininet source code may be found at https://github.com/mininet/mininet. We welcome new feature implementations, bug fixes, bug reports, FAQ additions and other contributions, and we look forward to high-quality discussions on mininet-discuss.

Thanks to all the code contributors, bug reporters, and active users who helped make this release possible!

The Mininet Team

  • Bob Lantz
  • Brandon Heller
  • Nikhil Handigol
  • Vimalkumar Jeyakumar

Website Updated

The old website had an obscure, needed-Google-to-find-it address, and was generated server-side, so it could be a bit slow. The new page you’re seeing was built with Octopress and is hosted on GitHub Pages. Since it’s all statically generated pages now, it should load super-quick, and it’ll be easier to add blog entries and do page customization.

Try it out, and let us know what you think!

-The Mininet Team

[Ed.: Apologies for use of the dubious term “website,” rather than the perfectly adequate “web site.” Perhaps you found us via Google or your favorite “searchsite.”]