IEEE Communications Magazine - June 2017 - page 220

IEEE Communications Magazine • June 2017
218
0163-6804/17/$25.00 © 2017 IEEE
A
bstract
The contemporary Internet has evolved from
an academic infrastructure to a tremendous
commercial network, serving as an indispensable
information platform for human communications,
but with inherent limitations such as complicated
management and manual configuration. To over-
come these limitations, the basic Internet archi-
tecture has unleashed an unprecedented wave of
innovation over the past decade, and introduced
a promising networking paradigm, software-de-
fined networking (SDN), often referred to as a
“radical new idea in networking.” SDN offers
numerous potential benefits such as enhanced
configuration, improved performance, and
encouraged innovation in network architectures
and operations, while undergoing unprecedented
challenges due to the lack of global network-wide
information. Nowadays, the emerging data-driven
thought opens the era of the fourth paradigm for
science research and transfers the design philoso-
phy of future networks, based on the large-scale
data rather than the small-scale data. We argue
that the existence of a data-driven information
plane in SDN can address these challenges. In this
article, we explicitly propose a data-driven infor-
mation plane and extend the SDN architecture
specifically. We first overview the evolution of
SDN, and then illustrate a novel SDN paradigm by
introducing a data-driven information plane into
the commonly accepted SDN reference architec-
ture. Finally, we outline emerging challenges and
discuss future directions for further research trig-
gered by a data-driven information plane.
I
ntroduction
The contemporary Internet has evolved from
an academic experimental infrastructure to a
tremendous commercial network, serving as an
indispensable information platform for human
communications. In a variety of network applica-
tions and servers, the Internet exposes its inher-
ent shortages such as complicated management,
manual configuration, untrusted servers, and hin-
dering innovation and evolution of the network-
ing infrastructure. As a result, the basic Internet
architecture has unleashed an unprecedented
wave of innovations. Over the past few decades,
a number of novel Internet architectures, such
as information-centric networking (ICN) [2] and
software-defined networking (SDN) [3-8], have
been proposed. These Internet architectures can
fall under two categories: incremental design and
clean-slate design [1]. Incremental design aims to
renovate the shortages of the current Internet,
without introducing any fundamental changes
to its basic underlying architecture. In contrast,
the clean-slate design, as a promising networking
design paradigm, promises to redesign the Inter-
net architecture and enable radical innovation
and evolution to overcome the limitations of the
Internet [1-2].
Among the proposed clean-slate architectures,
SDN as a “radical new idea in networking” has
achieved great triumphs from both the academic
community and industry, and has experienced
strong support by major Internet providers and
standardization groups. Today, the primary net-
working vendors are releasing network infrastruc-
tures to support SDN [4]. Furthermore, many
of them, such as Google and Microsoft, have
deployed SDN to interconnect their data centers
or public cloud worldwide [6]. SDN has provid-
ed numerous benefits ranging from centralized
network management and programmability to
reduced capital and operational expenses for the
current networks. The brain of SDN is centralized
in software-based controllers, which maintain a
global view of the network. Unfortunately, so
far, it has not been fully resolved how SDN can
become more intelligent, flexible, and powerful
with the global view of network-wide information.
To achieve this goal, an emerging burning issue
to be addressed is how to obtain the global view
of network-wide information from large-scale net-
work data rather than small-scale measurement
data or statistical data [14, 15].
Nowadays, it is an information-innova-
tion era, transitioning from a hypothesis-driven
world to a data-driven world stimulated by big
data, described as the fourth research paradigm
[14]. The ever-expanding massive network data,
which comes from countless network devices in
the form of device logs, usage histories, media
content delivered over networks, etc., reveals
remarkable insights into network design. In fact,
the networks have unleashed an unprecedented
wave of innovation with the emerging data-driv-
en idea, which changes the design philosophy of
future networks in almost all aspects, including
architecture design, resource management, and
Data-Driven Information Plane in
Software-Defined Networking
Haojun Huang, Hao Yin, Geyong Min, Hongbo Jiang, Junbao Zhang, and Yulei Wu
A
ccepted
from
O
pen
C
all
The authors propose a
data-driven information
plane and extend the
SDN architecture specifi-
cally. They first overview
the evolution of SDN,
and then illustrate a
novel SDN paradigm by
introducing data-driven
information plane into the
commonly accepted SDN
reference architecture.
They also outline the
emerging challenges and
discuss the future direc-
tions for further research
triggered by data-driven
information plane.
Haojun Huang is with Wuhan University; Hao Yin is with Tsinghua University; Geyong Min and Yulei Wu are with the University of Exeter;
Hongbo Jiang is with Huazhong University of Science and Technology; and Junbao Zhang is with Zhongyuan University of Technology.
Digital Object Identifier:
10.1109/MCOM.2017.1600114
1...,210,211,212,213,214,215,216,217,218,219 221,222,223,224,225,226,227,228
Powered by FlippingBook