Survey of software components to emulate OpenFlow protocol as an

American Journal of Software Engineering and Applications
2014; 3(6): 74-82
Published online December 16, 2014 (
doi: 10.11648/j.ajsea.20140306.12
ISSN: 2327-2473 (Print); ISSN: 2327-249X (Online)
Survey of software components to emulate OpenFlow
protocol as an SDN implementation
Mohammed Basheer Al-Somaidai, Estabrak Bassam Yahya
Dept. of Electrical Engineering, Mosul University, Mosul, Iraq
Email address: (M. B. Al-Somaidai), (E. B. Yahya)
To cite this article:
Mohammed Basheer Al-Somaidai, Estabrak Bassam Yahya. Survey of Software Components to Emulate OpenFlow Protocol as an SDN
Implementation. American Journal of Software Engineering and Applications. Vol. 3, No. 6, 2014, pp. 74-82.
doi: 10.11648/j.ajsea.20140306.12
Abstract: Software Defined Networks (SDN) is the next wave in networking evolution. It may be considered as a
revolution rather than an evolution since; many concepts of conventional network protocols are reshaped. OpenFlow protocol
is the most widely deployed protocol in SDN. Emulation of OpenFlow based network projects facilitates the implementation
of new ideas and driving the development of the protocol. In this paper, a summary of many software components related to
OpenFlow is presented. Most of these software components were tested by the researchers in order to simplify the choice for
other researchers considering the implementation of OpenFlow projects. These tests showed that there are differences in
performance for the controllers that support OpenFlow 1.0 and OpenFlow 1.3. Furthermore, the tested controllers differs in
the applications they support.
Keywords: Software Defined Network, OpenFlow, Emulation, Mininet
1. Introduction
A new paradigm in the field of networking is the software
defined networks a promising architecture, which is gaining
rapid attention of researchers and vendors as well [1-3]. This
is so because; the unlimited development of network
applications and the extensive demands of an explosive
growth in network users are driving conventional network
devices to their limits. SDN introduces a new way to handle
the vast amount of packets traversing the network. Many
packets belong to a single flow; thus, handling that flow and
distributing the actions to be taken to all its packets would
numerously speed up their forwarding. This is only one of
many other benefits of a centralized control of the network.
The most widely deployed SDN architecture is the OpenFlow
protocol. Many gigantic Internet vendors including Google
are considering the application of OpenFlow protocol in their
data centers [4]. A gradual implementation of SDN and
OpenFlow suggests the co-existence of OpenFlow networks
with conventional networks. This requires extensive studies
and projects to investigate the limitations and possibilities of
these protocols.
Simulation and emulation of network projects provide a
solid base to determine their pros and cons. Emulation is
more realistic than simulation since, it must be carried out in
real time and could provide a way to some real devices
running real operating systems to interact with some
simulated devices [5].
B. Lantz, et. al. [6] analyzed the performance of Mininet
emulator to develop, interact with, and customize the SDN
concept with OpenFlow protocol. This study showed
Mininet ease of use, scalability, and limitations.
S. Wang, et. al. [5] introduced the EstiNet OpenFlow
network simulator and emulator, and studied its performance
to design SDN networks. They compared EstiNet behavior,
capabilities and scalability with Mininet and ns-3 platforms.
A. Shalimov, et. al. [7] proposed a method to test and
compare popular open source SDN/OpenFlow controller.
They analyzed throughput, latency, scalability and security
by developing new framework called Hcprobe based on
Cbench framework.
B. Nunes, [8] provided historic review about
programmable network idea from its beginning time down
to the SDN revolution. The study presented the architecture
of SDN and discussed OpenFlow features, application and
related software to deploy and develop SDN networks based
on OpenFlow.
A. Lara, et. al. [9] discussed the architecture of OpenFlow
Mohammed Basheer Al-Somaidai and Estabrak Bassam Yahya: Survey of Software Components to Emulate OpenFlow
Protocol as an SDN Implementation
network to understand SDN, and centralized control concept
by using different controllers' platform. In addition, studies
have measured the performance of OpenFlow networks
through modeling and experimentation. The researchers
clarify the challenges facing the large-scale OpenFlow
networks and applications.
The rest of this paper is structured as follows: in section 2
we briefly discuss software defined network architecture.
Section 3 introduces an overview of OpenFlow protocol, its
fundamental concepts and messages. Some SDN and
OpenFlow platforms are presented and compared in section
4; while, sections 5 and 6 give a survey of OpenFlow
software controllers and switches respectively. Any
OpenFlow project could make use of some tools that are
presented in section 7. Finally, section 8 contains some
concluding remarks and future work suggestions.
2. Software Defined Networks (SDN)
The adaptation of packet switching in networking made
each network device such as gateway, router, or switch a
standalone device. These devices manage themselves
independently even if this management was according to a
certain routing protocol or administration policy. Each data
packet undergoes the same parsing and processing efforts at
each network node even if it belongs to the same flow. This
conventional architecture of networks may fail to support
the dramatically increase in users' requirements and the fast
deployments of new network applications.
Segregating the control plane and the management plane
from the data forwarding plane in network devices is what
software defined network (SDN) about [10,11]. In such a
paradigm, a central controller is responsible for managing
many forwarding devices that lay under its supervision. Such
configuration would results in efficient, faster innovative, and
more scalable networks that meet users' demands. Software
defined network is managed through a network operating
system implemented at the controller to make all the
subsequent switches work in harmony and more flexibility.
These switches need not be in the same geographical area;
the management of many planet wise distributed data
centers that belong to a cloud service provider is an example
of this diverse distribution of forwarding devices [2]. Fig. 1,
shows the architecture of a software defined network. It is
worth to mention that SDN is not a protocol; but it is an
operational and programming architecture. Albeit, SDN
uses certain protocols for making the network
programmable. These could be OpenFlow [12], I2RS,
PCE-P, BGP-LS, NetConf/Yong, and OMI [11]. In this paper,
we are focusing on the widely deployed OpenFlow protocol.
controller, an OpenFlow switch, and the OpenFlow protocol.
The Open Networking Foundation (ONF) a non-profit
Fig 1. Architecture of a software defined network.
organization was created in 2011 by a group of vendors [14].
It is dedicated to coordinating the development of SDN
standards and solutions in order to accelerate the delivery of
SDN products, services, and applications. Since then ONF
had published each new version of OpenFlow standard. Up to
the date of writing this paper (March 2014) the last version of
OpenFlow switch specification is 1.4 and it was published in
October 2013 [12]. According to this specification, the
architecture of an OpenFlow switch should contain the blocks
shown in Fig. 2, each OpenFlow switch contains one or more
flow tables processed in pipeline, a single group table, a single
meter table; and a various types of ports. Each table and port
in the OpenFlow switch is associated with many counters that
could gather various statistics describing the events that the
switch is subjected to. The controller creates all the tables and
their entries; the data packets that traverse the OpenFlow
switch update the counters.
The corner stone in the OpenFlow protocol is the flow
table, which has 256 entries. Each entry in the flow table
contains six sections as shown in Fig. 3.
3. Open Flow
OpenFlow started at Stanford University in 2008 [13].
The aim of the project was to give researchers a tool to
implement their experimental protocols in networks.
OpenFlow network consists of three major components: a
Fig 2. Architecture of an OpenFlow switch
American Journal of Software Engineering and Applications 2014; 3(6): 74-82
Fig 3. OpenFlow switch flow table entry fields.
The matching fields section is used to match the packet
with the entry according to various packet header fields.
When more than one entry match a packet the priority field
determines the flow table entry that will be executed and the
per flow table entry counters are updated.
The instructions section contains among other things the
actions that will be acted upon the matched packet. The
timeouts field specifies the maximum amount of hard time
and idle time before the flow table entry expires. A zero
value in any of them disable the corresponding timer. The
hard timeout determines the maximum amount of time in
seconds before the flow table entry expires; while the idle
time out causes the expiration of the entry if it has matched
no packet in the given number of seconds. The cookie field
is used by the controller to filter flow statistics, flow
modification, and flow deletion. Each flow table must
support a table-miss flow entry clarifying the action that
should be taken upon the unmatched packet either sending it
to the controller, dropping it, or directing it to the subsequent
flow table in the pipeline [12].
OpenFlow protocol has three types of messages to
communicate between the controller and the OpenFlow
switch over a secure channel or over a TCP channel as
shown in Fig.4. They are classified according to the initiator
of the message into controller to switch messages,
asynchronous (switch to controller) messages, and
symmetric messages. The controller to switch messages are
used to assert its control upon the switch, reading the switch
status, and modifying the switch states which includes
editing the switch flow tables.
The switch to controller messages are used to inform the
controller about a new incoming flow, a change in a switch
state; or a request for modifying a flow table entry. Either the
controller or the switch could initiate the symmetric
messages. They include hello messages, echo messages,
error messages, and experimenter message that identify the
vendor of the controller or the switch [12]. Table 1. shows a
summary of OpenFlow switch standards specification
properties. It can be observed that almost every year there is
a new version in the 1.x numbering of the standard, and
although OpenFlow protocol is still in its 1.x version, there
is huge development every year.
4. SDN Development Platforms
There are many platforms that could be used by
researchers to emulate and/or simulate their SDN projects.
Researchers use these tools to perform experiments, study
the behavior of the network, and develop new methods to
support different applications. In this section, a description
of these currently available SDN platforms is presented
emphasizing on the rapidly developed and deployed Mininet
platform. Table 2. gives an overview of some properties of
these platforms.
Fig 4. OpenFlow protocol messages.
Table 1. OpenFlow switch standards properties.
Version / Property
Publication date
Dec. 31, 2009
Feb. 28, 2011
Dec. 5, 2011
Jun. 25, 2012
Oct. 15, 2013
Widely deployed
Flow table
Group table
Meter table
IPv6 support
Stand alone
secure mode
Stand alone / secure
Stand alone /
secure mode
Stand alone
secure mode
Multiple controller
Eviction /Vacancy/Synchronization
Optical ports
Controller connection failer
Mohammed Basheer Al-Somaidai and Estabrak Bassam Yahya: Survey of Software Components to Emulate OpenFlow
Protocol as an SDN Implementation
Table 2. Properties of SDN platforms
Last version
Stanford University,
ON. Lab
EstiNet Technologies Inc.
ns-3 Project
NEC Corporation
Web site
www.mininet. org
www.estinet. com
io/trema/ Trema.github.
Operating system
Ubuntu, Fedora
Fedora (14,17)
Windows, FreeBSD, Mac,
GNU/ Debian, Ubuntu,
OpenFlow versions
1.0 – 1.3
1.0, 1.1, 1.3, 1.3.2
1.0, 1.3, 1.3.1
VND*, Miniedit
EstiNet GUI
Emulation mode
Simulation mode
Free / Proprietary
*VND: Visual Network Description, to be mentioned in section 7
4.1. Mininet
Mininet is a network emulation platform that supports
rapid development in SDN using OpenFlow protocol. It is
the most popular SDN platform used by SDN researchers
due to its simplicity, availability, and flexibility. Furthermore,
Mininet is entirely devoted to OpenFlow architecture [6].
Mininet uses Linux kernels along with Python language
scripts to construct a virtual network of large number of
hosts network, OpenFlow switches, and controllers in any
network topology the researcher employs over a single
desktop or laptop station.
Mininet could use its built-in software tools to develop
such networks through Command Line Interface (CLI), or
adapts to a third-party software tools that implement other
controllers or Graphic User Interface (GUI) engines [15, 16].
It has the flexibility of adding many controller types that will
be mentioned in section 5.
4.2. EstiNet
EstiNet is an emulation and simulation platform of many
network protocols; one of them is OpenFlow protocol. It
also supports some of the controllers of section V. EstiNet is
a proprietary software tool and it uses the company servers
to run the simulation or the emulation projects. This cloud
service is referred to as Simulation as a Service [17].
EstiNet has good simulation properties among them are
accurate and repeatable result with a graphical user interface
and packet animation along with good presentation of the
simulation statistics as a graph for each node in the network [5].
4.3. ns-3
ns-3 is a well established network simulator usually
compared to OPNET for providing simulation environment
to a wide range of network protocols. ns-3 supports
OpenFlow protocol and its switches in simulator
environment but it cannot readily run a real OpenFlow
controller such as NOX, POX, or Floodlight without
modifications. This is why ns-3 has implemented its own
OpenFlow controller as a C++ module with a different
performance from the above real controllers.
Another drawback of using ns-3 is that it until now
supports version 0.89 of OpenFlow protocol only, this limits
the researchers' ability to test and develop projects that are
compatible with the new versions of OpenFlow protocol [5].
It could be used to introduce the concepts of SDN and
OpenFlow to beginners who are used to ns-3.
4.4. Trema
Trema is an OpenFlow framework that includes
everything the researcher needs to conduct an OpenFlow
project. The source tree includes basic libraries and
functional modules that work as an interface to OpenFlow
switches. Several examples of sample applications are also
It has an integrated testing and debugging environment
that manage, monitor, and diagnose the entire system with a
network emulator and a diagnostic tool chain (Trema shark,
Wireshark plug-in) [18]. The lack of a graphical user
interface and the use of the programming languages C and
Ruby may limit the popularity of this platform.
5. Controller Software
A block diagram of the controller; which is the brain of
any software defined network is shown in Fig. 5. The
controller communicates with the forwarding devices
through an SDN protocol such as OpenFlow. This link is
also called the southbound Application Programming
Interface (API). From the other side the controller uses a
northbound API to deal with various applications. If we
made an analogy for the network as an orchestra then the
controller plays the role of the maestro. In fact, some SDN
implementations use these designations to refer to SDN and
the controller [11, 19].
As the basic concept of SDN is to decouple the control
American Journal of Software Engineering and Applications 2014; 3(6): 74-82
plane and the management plane from the data-forwarding
plane then the controller has to bear all the burden of
controlling and managing all the data forwarding devices. It
should maintain and update through the rule-placement
algorithm information about all the forwarding devices that
are under direct responsibility of the controller including
their flow tables, links, and states. The routing policy is
another task of the controller any change in any forwarding
device state causes the controller to reshape the routing path
of all flows traverse that device resulting in updates to a
large number of switches' flow tables. Security strategies
along with end devices policy are also, placed in the
switch feature [20]. NOX combined with Mininet provides
a platform for academic research in networking [21]. It
supports now many features of OpenFlow protocol
specification 1.3, but the researchers when implement this
version discovered the Iperf command which determine the
bandwidth utilization does not work properly.
5.2. POX
POX controller is another SDN control platform and it is
considered an active development tool. POX was derived
from NOX controller platform with the main difference is
using Python programming language instead of C++
platform. POX uses Python API (version 2.7) to support
network virtualization, SDN debugging, and different
application such as layer-2 switch, bridge, hub, etc [22].
NOX and POX controllers support the same GUI and
visualization tools to setup, configure controllers, and flow
tables. POX still does not support OpenFlow 1.3, which
many other controllers support now.
5.3. Floodlight
Fig 5. Block diagram of the controller.
As mentioned above the controller plays a vital role in the
OpenFlow network; therefore multi controllers could
establish communication with a forwarding device (switch)
provided that only one of them has the master role upon the
switch and the others should be in the slave role. Having
multiple controllers improves reliability, as the switch can
continue to operate in OpenFlow mode if one controller or
controller connection fails. The hand-over between
controllers is entirely managed by the controllers themselves,
which enable fast recovery from failure and controllers load
balancing [12]. Many software implementations of the
controller are summarized in Table 3.
5.1. NOX
NOX controller was the original controller of OpenFlow.
It is written in C++ language and its first version provided an
API for Python scripts, but last version of NOX has dropped
this API and supported C++ only. NOX provides a
high-level programmable interface upon forwarding devices
and applications. It is designed to support both small
networks of a few hosts and large enterprise networks of
hundreds of switches and hosts.
NOX's core has features of fast, asynchronous I/O,
topology discovery, host tracking possibility, and learning
Floodlight is a very popular SDN controller. It is a
contribution from Big Switch Networks and it uses Java
based platform (API) thus it runs within a Java Virtual
Machine (JVM) and it is considered suitable with
continuous increase in number of network devices (switches)
that deal with OpenFlow concept [11,23].
Floodlight controller realizes a set of common
functionalities to control and inquire an OpenFlow network.
The controller has features of simple to extend and enhance,
easy to setup with minimal dependencies, support for Open
Stack Quantum cloud, topology management, and it deals
with mixed OpenFlow and non-OpenFlow network.
Floodlight supports applications that include a learning
switch, hub application, firewall, and static flow push
applications [21]. Floodlight as POX does not support
OpenFlow 1.3.
5.4. OpenDaylight
OpenDaylight is an OpenFlow controller. It has open and
reference framework for programmability and control
through open source SDN, it uses JVM so it can be used with
any platform or operating system that supports Java 1.7+. It
is a modular, extensible, scalable and multi-protocol
controller infrastructure built for SDN deployment on
modern heterogeneous multi-vendor networks [21, 24].
OpenDaylight enables users to reduce operational
complexity, extend the lifetime of their testing network
infrastructure, and enable new services and capabilities. In
our test of OpenDaylight, it proved to have an excellent GUI,
but Iperf command undergo the same problems that we
faced with NOX when dealing with controller.
Mohammed Basheer Al-Somaidai and Estabrak Bassam Yahya: Survey of Software Components to Emulate OpenFlow
Protocol as an SDN Implementation
Table 3. Controller software implementations.
OpenFlow versions
Operating system
C ++
1.0, 1.3
Linux, Windows,
Big Switch Networks
Floodlight web UI, Avior
Linux, Mac
Linux Foundation
Collaborative Project
1.0, 1.3
OpenDaylight web UI
Linux, Windows
Nippon Telegraph and
Telephone Corporation
1.0, 1.2, 1.3 and
Nicira extension
1.0, 1.3.1
Stanford University
Windows, Linux,
5.5. Ryu
Ryu is a component-based, open source framework
implemented entirely in Python. Nevertheless, the Ryu
messaging service does support components developed in
other languages [25].
The goal of Ryu is to develop an operating system for
SDN that has high quality enough for use in large networks.
Ryu controller includes event management, in-memory
state management, application management, and series of
reusable libraries (e.g NetCOONF library, sFlow/NetFlow
library and OF-Config library). Additionally, it supports
applications such as OpenStack Quantum, layer-2 switch,
Generic Routing Encapsulation tunnel interface (GRE), and
tunnel abstractions. As well, as services about topology and
statistics [11].
5.6. Mul
Mul is an OpenFlow SDN controller and it uses C based
multi-threaded infrastructure at its core and it is designed to
provide good services and ensure reliability through the
network [26]. Mul supports OpenFlow 1.3.1and did not
work in our test with OpenFlow 1.3 switches such as Open
5.7. Beacon
Beacon is an OpenFlow SDN controller and it uses Java
based API. Beacon has features of rapid development, fast
and dynamic performance in order to code bundle features
5.8. Special Purpose Controllers
There is a type of controllers; that operates with general
purpose controllers such as FlowVisor, and RouteFlow [21].
FlowVisor acts as a proxy between an OpenFlow switch and
multi controllers. So that it directs the first packet of a new
flow to the appropriate controller according to application,
port, MAC, or IP address. This would results in the
separation of the network or applications into slices where
each slice is controlled by a different controller [28]. It does
not support OpenFlow 1.3 yet.
RouteFlow can be considered as a network application on
top of general OpenFlow controllers. The major objective of
RouteFlow is to build up an open source framework for
virtual IP routing solution over product hardware
implementing the OpenFlow API [29].
6. Switch Software
OpenFlow switch is an important component of software
defined network, switch connects with controller and when a
packet arrives to the switch; the switch performs a number of
processes, compares the packet header with flow entries, and
identifies the actions to be implemented as illustrated in
prior sections. Mininet can support different type of switches
such as:
6.1. Open vSwitch (OVS)
Open vSwitch is a production quality open source
software switch designed to be used as a virtual switch in
large scale virtualized environments. Open vSwitch supports
many flavors of Linux operating systems such as Debian,
Ubuntu, and Fedora. Furthermore, it supports Windows and
FreeBSD operating systems [30].
Open vSwitch uses OpenFlow protocol to support the
efficient management, virtual switch configuration, and QoS
policies need to be applied across a large number of hosts.
Open vSwitch supports OpenFlow versions 1.0, 1.1, 1.2, 1.3.
As well, it supports other standard management protocols
such as SNMP or NETCONF. Additionally, Open vSwitch
provides interfaces to monitoring protocols such as sFlow
and NetFlow [31]. Open vSwitch is commonly used with
Mininet emulator for testing networks that use OpenFlow
protocol [21].
6.2. OFSoftSwitch13
OFSoftSwitch13 is an OpenFlow 1.3 compatible
user-space software switch implementation. This project is
American Journal of Software Engineering and Applications 2014; 3(6): 74-82
supported by Evicsson Innovation center/Brazil [21].
Mininet users can install the switch software, NOX
controller that supports OpenFlow version 1.3, and
download useful documentation to run and configure
OFSoftSwitch13 from public Github web site [32].
6.3. LINC
LINC is an open source project that supports OpenFlow
protocol versions 1.2, and 1.3. LINC is architected to use
generally available commodity, x86 hardware and runs in
various operating systems such as Linux, Windows, Mac,
etc [21]. Mininet user can install this switch software from
Github web site [33].
6.4. Indigo Virtual Switch (IVS)
Indigo project is an open source project, which supports
OpenFlow protocol on physical and hypervisor switches. It
is designed for high performance and minimal
administration and it uses the hardware feature of
Application Specific Integrated Circuit (ASICs) of Ethernet
switch to run OpenFlow at line speed [21].
Indigo Virtual Switch is a lightweight high performance
virtual switch support OpenFlow version 1.0 only. It is
designed to enable virtualization in big networks
applications as it is used with floodlight controller [34].
7. Tools
Mininet emulator can be integrated with a number of open
source tools to meet and implement the different needs of
Mininet users, such as: editors, GUI, and benchmarks, ..etc.
7.1. Editors
Mininet user can use one of the Integrated Development
Environment (IDE) supported by Mininet environment such
as Python IDLE version 2.7, Python IDLE version 3.2, GNU
Emacs editor, and Nano editor as a text editor for writing
code to build and configure the network topology.
7.2. Graphic User Interface (GUI)
There are a number of GUIs that are used to configure
network elements (controller, switches, and hosts) and
display network topology. They include many component
such as:
7.2.1. Miniedit
Miniedit is a simple Python script presented with Mininet
examples. It is used as GUI to construct network topology
and emulate it.
Miniedit was developed to add new features and
capabilities for the purpose of forming a networks, such as
the use of the remote controller and multi controllers, select
properties of the links, controller, switches, and hosts,
provide command line interface terminals for each node, use
monitoring protocols (sFlow, NetFlow), and export python
script for network topology [15].
7.2.2. Visual Network Description (VND)
Visual Network Description-SDN version is an online
GUI used to form network topology and configure node
properties, link type and properties, setup switches flow
table entries, and export network topology and its
configuration as a Python script to Mininet emulator and
OpenFlow controllers or as a C++ script to ns-3 simulator
7.2.3. Avior
Avior is a GUI used with floodlight controller. It provides
features of eliminating dependency on using Python script
and API in order to manipulate network and monitoring its
behavior [35].
Avior has flow manager tools and could give a summery
about controllers, switches, and hosts. Controllers summery
provides information about host names, JVM memory bloat
and other controllers information. Switches summery
provides information about port, counters, match header
fields, and switch flow entries (add/ delete). On the other hand,
host summery provides information about the attached switch
Data Path ID (DPID) and the switch port connect to it [21].
7.2.4. Web User Interface (UI)
It is one of the GUI used with some controllers such as
Floodlight, and OpenDaylight. It is an online GUI; where user
can access it after installing and running the controller using
the URL address (http://localhost:8080). Web UI displays
topology of network run in Mininet, network node (switches,
hosts) information such as IP, MAC, and DPID, flows outline
and add/remove switches flow tables entries,..etc.
7.3. Benchmarks
In order to test network performance, many benchmarks
could be used. The following are examples for benchmarks.
7.3.1. OFtest
It is a framework and collection of testes for validating
OpenFlow switches. OFtest provides a connection as a
controller to the OpenFlow switch and send messages to test
OpenFlow basic functionalities. It supports OpenFlow
specification versions (1.0-1.3) [21].
OFtest uses Python and Scapy as a pre-requisites, where
Scapy is a powerful interactive packet manipulation
program; used to decode packets, match requests with
replies. It also can handle tasks like scanning, and trace
routing [36].
7.3.2. OFlops
It is an OpenFlow testing platform used to focus on
OpenFlow protocol behavior by implementing basic
measurement tests that allow developers to specify and
study the capabilities of OpenFlow devices [37].
OFlops tests are used to assess performance of OpenFlow
switches in network by utilizing multi-threading parallelism
[21]. OFlops has features of modularity, low overhead with
minimum delay in processing to support parallelism, and
heterogeneity by being compatible with a number of packet
Mohammed Basheer Al-Somaidai and Estabrak Bassam Yahya: Survey of Software Components to Emulate OpenFlow
Protocol as an SDN Implementation
generation and capturing tools such as Cbench and
7.3.3. Cbench
It is a program for testing OpenFlow controllers by
generating packet-in messages and waits for flow-mods
messages to receive. Cbench has two emulated modes:
latency mode and throughput mode. Cbench can be used to
measure controller performance by changing its arguments
such as number of switches, number of MACs per switch
(hosts), number of tests and time of test [7]. Cbench
supports OpenFlow 1.0 only, but the Mul controller vender
Kulcloud introduced a modified version of Cbench that
supports OpenFlow 1.3 and is called Kcbench, albeit, it
worked with Mul controller only in our test.
7.4. Linux Kernel Programs
Because Mininet emulator uses Linux kernels, it supports
a number of Linux programs and commands such as Dump,
Ping, Pingall, Iperf, and plot programs like Gnuplot
program; which supports many types of plots in 2D and 3D.
The Dump command illustrates network nodes with their
interfaces connections. Ping and Pingall test network
connectivity and latency. Iperf determines bandwidth
utilization and retransmission of packets in TCP
applications. It also measures loss and jitter for UDP
related to OpenFlow protocol. Most of them were
downloaded, installed, and operated successfully.
OpenDaylight, Floodlight, and OFSoftSwitch13 proved to
have good properties like good documentation and flexibility.
Observing the rapid development of OpenFlow standards
predicts that a major breakthrough is expected in version 2.0,
but for the time being the use of software components that
supports version 1.3 like NOX, OpenDaylight, and Mul is
recommended since no software component supports the new
1.4 version yet.
Most of the tested software components are standalone
components. The need for a frame that gather the installation
and operation of switches and controllers into a single
platform with a certain GUI and benchmark would facilitate
the development of OpenFlow projects. EstiNet is a good
example of such a platform. An emulation projects to test the
compatibility of OpenFlow protocol with WLAN and IPv6
deployment is under consideration by the researchers.
M. Mendonca, B. Nunes, K. Obraczka, and T. Turletti,
"Software defined networking for heterogeneous networks,"
IEEE COMSOC MMTC E-Letter United States, vol. 8, no. 3,
pp. 36-39, May 2013.
S. Jain, A. Kumar, S. Mandal, J. Ong, L. Poutievski, A. Singh,
S. Venkata, J. Wanderer, J. Zhou, M. Zhu, J. Zolla, U. Holzle,
S. Stuart and A. Vahdat, "B4: experience with a
globally-deployed software defined WAN," SIGCOMM’13
Hong Kong, China, pp. 3-14, August 2013.
D. Drutskoy, E. Keller, and J. Rexford, "Scalable network
virtualization in software-defined networks," IEEE Internet
Computing, vol.17, pp. 20 – 27, March-April 2013.
Migration Working Group, "Migration use cases and
methods," Open Networking Foundation (ONF), February
S. Wang, C. Chou, and C.Yang, " EstiNet OpenFlow network
simulator and emulator," IEEE Communications Magazine,
vol. 51, pp. 110-117, September 2013.
B. Lantz, B. Heller, and N. McKeown, "A network in a
lap-top: rapid prototyping for software-defined networks," In
Proceedings of the 9th ACM SIGCOMM Workshop on Hot
Topics in Networks New York,2010.
A. Shalimov, D. Zuikov, D. Zimarina, V. Pashkov, and R.
Smeliansky, "Advanced study of SDN/OpenFlow
controllers,"CEE-SECR'13 Proceedings of the 9th Central &
Eastern European Software Engineering Conference in
Russia, November 2013.
B. Nunes, M. Mendonca, X. Nguyen, K. Obraczka, and T.
Turletti, "A survey of software-defined networking: past,
present, and future of programmable networks," IEEE
Communications Surveys & Tutorials, in press, January
A. Lara, A. Kolasani, and B. Ramamurthy, "Network
survey," IEEE
Communications Surveys & Tutorials, vol. 16, pp. 493 – 512,
February 2014.
7.5. Frenetic
Frenetic is a domain-specific language used to program
software defined networks [38]. It has features of high-level
abstractions. Therefore, it is useful to replace the low-level
interfaces available today. Frenetic offers a suite of
information about network state, identity, forwarding
policies, and updating policies [39].
7.6. Wireshark
Mininet supports Wireshark packet analyzer and uses it to
capture packets traverse the network nodes and analyze
these packets to study performance of the network and
obtain statistical measurements about its behavior [40].
OpenFlow messages could be displayed and studied using
Wireshark. Wireshark version 1.11 and above supports a
new filter for OpenFlow 1.3 packets.
8. Conclusions and Future Works
Many SDN protocols are available now, but employing the
OpenFlow protocol is highly recommended due to its open
source nature, rapid development, and wide deployment.
The proper use of emulation software components in
developing OpenFlow and SDN projects would save a lot of
time and money compared to practical testbeds since real
hardware devices are still expensive and support primitive
versions of OpenFlow standard only.
In this paper, we examined many software components
American Journal of Software Engineering and Applications 2014; 3(6): 74-82
[10] Open Network Foundation (ONF), "Software-defined
networking: the new norm for networks," April 2012.
openflow/wp-sdn-newnorm.pdf, accessed on 23/3/2014.
[23] POX,
[11] T. Nadeau and K. Gray, SDN: Software Defined Networks,
1st ed., O’Reilly Media, Inc., August 2013.
[26] Ryu,
[12] Open Network Foundation (ONF), "OpenFlow switch
specification, version 1.4.0 (wire protocol 0x05)," October
2013. Available at:
[28] Beacon,
4.0.pdf, accessed on 23/3/2014.
[24] Floodlight,
[25] OpenDaylight,
[27] Mul,
[29] FlowvisorExecise,
[30] RouteFlow,
[14] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L.
Peterson, J. Rexford, S. Shenker, and J. Turner, "Openflow:
enabling innovation in campus networks," SIGCOMM
Comput. Commun. Rev., vol. 38, no. 2, pp. 69–74, March
[31] Open vSwitch,
[15] Open Networking Foundation (ONF),"Open networking
foundation," Available at
[33] OpenFlow
[16] Tech
[34] LINC,
[17] Visual Network Description (VND), available at:
[18] EstiNet Technologies Inc., "The GUI user Manual for the
EstiNet 8.0 Network Simulator and Emulator", January 2013.
[19] T. Dietz, "Trema tutorial," NEC Corporation, March 2012.
utorial.pdf, accessed on 23/3/2014.
[20] Z. Cai, A. Cox, T. Eugene Ng, "Maestro: balancing fairness,
latency and throughput in the OpenFlow control plane," Rice
University Technical Report TR11-07, 2011.
[21] NOX,
[22] S. Azodolmolky, Software Defined Networking with
OpenFlow, Packt Publishing, 1st ed., October 2013.
[32] A. Clemm, and R. Wolter, "network-embedded management
and applications understanding programmable networking
infrastructure," Springer New York, 2013.
[35] Indigo,
[36] Avior,
[37] Scapy,
[38] OFLops:
[39] Frenetic,
[40] N. Foster, M. Freedman, A. Guha, R. Harrison, N. Katta, C.
Monsanto, J. Reich, M. Reitblatt, J. Rexford, C. Schlesinger,
A. Story, D. Walker, "Languages for software-defined
networks," IEEE Communications Magazine, vol. 51, pp.
128 – 134, February 2013.
[41] R. Shimonski, Analyzing and Troubleshooting Network
Traffic, Elsevier Inc., 1st ed., 2013.