F. Robust Wireless Integration
Although OpenFlow has been well developed in wired, local
network, there are very few studies on its performance in
wireless networks. All the existing wireless network OpenFlow
designs [19], [35], [111]–[114] have not overcome the follow-
ing two challenging issues.
1) Integrated Management of Channel Access and Data For-
warding: Many wireless networks support multi-channel com-
munications among routers and clients, especially in cognitive
radios. Therefore, each wireless node needs to detect available
channels, and select the high-quality ones. Moreover, the nodes
can change physical characteristics for optimal link radio com-
munications. Thus in a wireless networks the OpenFlow data
panel must perform efficient channel sensing/access. However,
the existing OpenFlow standards only define data forwarding
functions in the hardware. We will need to significantly improve
the existing OpenFlow model (including its control/data panels’
task division, FlowVisor control, network visualization, etc.) by
designing a brand-new flow-table management scheme. Such
a scheme may use reinforcement learning to simultaneously
manage two flow tables one for real-time data forwarding and
another for multi-channel sensing/selection.
2) Conflict-Free Scheduling of Control Traffic and Data
Traffic: Unlike wired OpenFlow model that can use cables
to easily achieve control/data packet communications among
nodes, wireless network uses unreliable wireless links for both
control panel communications and data panel packet forward-
ing. The control panel demands a high-quality channel for loss-
free delivery, while the data panel should use other available
channels for routing. Therefore, a carefully designed chan-
nel allocation and packet scheduling scheme is required for
conflict-free control/data traffic delivery in routers and nodes.
X. C
ONCLUSION
In this paper, we have comprehensively surveyed the design
issues for SDN/OpenFlow. Especially we have covered all
important issues in concrete network implementation including
language abstraction, controller design, virtualization scheme,
QoS support, security issues, and wireless/optical network
integration.
Below we briefly summarize some important aspects for
highlight: SDNs have many applications including developing
new protocols prior to implementing them in real networks,
increasing connectivity in rural environments, making both
cloud based and regular data centers better, and supporting
mobile device offloading. As the Internet continues to grow
and becomes available to increasing more people, networks will
need to be able to adapt to ever changing circumstances. SDNs
allow the data and control planes to be separated, and hence to
be easier for improvements.
Network virtualization based on OpenFlow is a successful
implementation of Software Defined Networking, and its de-
velopment offers users a great deal of convenient services. We
have reviewed different virtualization architectures that focus
on the improvement of the network flexibility, isolation and
management. It can be seen that embedding additional module
or abstraction layer on the top of OpenFlow or FlowVisor
provides solutions to these challenges. Utilizing the database
can also help to simplify the creation of the abstraction layer.
QoS is an import issue in many applications especially in
streaming media, VoIP, Videoconferencing, and so on. Many
experiments have been conducted to make OpenFlow support
QoS control. However, these designs are still under testing
phase, and need to be further examined. Many designs are
related to optimization problems, such as dynamic rerouting
for SVC, dynamic QoS re-negotiation for multimedia flows,
etc. And the solution needs heavy calculations in reality as the
dimension increases. It also needs a comprehensive test before
being applied to real world applications. Many experiments are
actually under a small scale of tests to verify the proposed de-
sign, and no large-scale experiments have been performed yet.
Security and privacy are always important in any network.
OpenFlow brings many new security challenges due to the
virtual network management. It is important to design new
low-overhead security/privacy schemes to protect the virtual-
to-physical mapping protocols in SDN/OpenFlow.
We have also pointed out some important unsolved research
issues in this exciting field. Those issues may serve as the
thesis/dissertation topics for graduate students. SDN/OpenFlow
is a relatively new field, and many practical design issues
are waiting for in-depth investigations. We believe that this
comprehensive survey could help R&D people to understand
the state-of-the-art in SDN/OpenFlow.
A
CKNOWLEDGMENT
We sincerely thank the following people for their help
with this survey: A. Gerrity, M. Farooq, T. Zhang, R. Ma,
C. Dickerson, X. Fu, N. Hegde, (they are all with ECE de-
partment at the University of Alabama), and A. V. Vasilakos
(University of Western Macedonia in Greece). They have pro-
vided valuable comments and inputs on some of the above
discussed topics. We also appreciate the editor and reviewers’
time and effort for reviewing this paper.
R
EFERENCES
[1] S. Ortiz, “Software-defined networking: On the verge of a break-
through?” Computer, vol. 46, no. 7, pp. 10–12, Jul. 2013.
[2] H. Kim and N. Feamster, “Improving network management with
software defined networking,” IEEE Commun. Mag., vol. 51, no. 2,
pp. 114–119, Feb. 2013.
[3] K. Bakshi, “Considerations for software defined networking (SDN):
Approaches and use cases,” in Proc. IEEE Aerosp. Conf., Mar. 2013,
pp. 1–9.
[4] S. Agarwal, M. Kodialam, and T. V. Lakshman, “Traffic engineering
in software defined networks,” in Proc. IEEE INFOCOM, Apr. 2013,
pp. 2211–2219.
2204
IEEE COMMUNICATION SURVEYS & TUTORIALS, VOL. 16, NO. 4, FOURTH QUARTER 2014
[5] S. Fang, Y. Yu, C. H. Foh, and K. M. M. Aung, “A loss-free multipathing
solution for data center network using software-defined networking ap-
proach,” IEEE Trans. Magn., vol. 49, no. 6, pp. 2723–2730, Jun. 2013.
[6] S. H. Yeganeh, A. Tootoonchian, and Y. Ganjali, “On scalability of
software-defined networking,” IEEE Commun. Mag., vol. 51, no. 2,
pp. 136–141, Feb. 2013.
[7] S. Das et al., “Packet and circuit network convergence with OpenFlow,”
in Proc. Opt. Fiber Commun. Conf. Expo., Mar. 2010, pp. 1–3.
[8] R. Sherwood, M. Chan, and A. Covington, “Carving research slices
out of your production networks with OpenFlow,” ACM SIGCOMM
Do'stlaringiz bilan baham: |