Performance optimisation of control channel
in ForCES-based software defined network
ISSN 2047-4954
Received on 17th February 2016
Revised on 26th July 2016
Accepted on 10th August 2016
doi: 10.1049/iet-net.2016.0009
www.ietdl.org
Chuanhuang Li
✉
, Liang Gong, Lijie Cen, Weiming Wang
College of Information and Electronic Engineering, Zhejiang Gongshang University, Hangzhou 310018, People’s Republic of China
✉ E-mail: chuanhuang_li@zjgsu.edu.cn
Abstract: The traditional network architecture cannot meet the growing needs of increasingly diverse network
applications. The new generation network architectures, software defined ne twork (SDN) which have features of high
openness, flexibility, scalability and high controllability have widely been studied. However, a seri es of new problems
are also produced in this new architecture, such as security problem resulted from the high openness, and
performance problem resu lted from the high flexibili ty. Currently, the researchers are mainly concentrated in the
implementation and standardisation of SDN. There is nogoodapproachtoevaluateandoptimisethesystem
performance. This studies the performance problems unresolved in SDN research area, and proposed an idea and
method to analyse the control channel of ForCES (Forwarding and Control Element Separation)-based SDN by using
the stochastic network calculus theory. First, the architecture of ForCES-based SDN is introduced. Then, the
performance model of the channel is given. Based on the model,anoptimisationmethodtoperformanceisdetailed.
This study also gives a simulation by using NS-2 (Network Simulator, Version 2) to verify the correctness and reliability
of the performance model.
1 Introduction
Software defined network (SDN) which is regarded as the ultimate
solution for the future network [1] has been given a special
attention in recent years. The goal of SDN is to flexibly control
the network and network nodes, in a centralised and uni fied
software method, and on multiple levels of abstraction. Today how
to implement a more comprehensive SDN is still a very
controversial issue. Although the concept of SDN was first
introduced by OpenFlow [2] technology, many companies such as
Google, Huawei, Cisco and IETF (Internet Engineering Task
Force) organisation generally believe that OpenFlow technology
lags far behind the realisation of SDN [3, 4]. Flow table definition
cannot efficiently define network resource slices [5], which greatly
limit the utilisation and function implementation. In 2012, even
ONF (Open Network Foundation) also set up a working group
(FAWG, Forwarding Abstractions Working Group) to try to come
up with better ways to more fully define the network resources and
to meet the application requirements [6].
In 2012, IETF set up a special research group for SDN (SDNRG,
SDN Research Group) [7] to propose implementatio n solutions by
integrating all possible IETF tech nologi es. In the current IE TF
technologies, the ForCES (Forwardin g and Control Element
Separation) technology [8] which can provide a powerful netwo rk
resource description method is paying attention by lots of
companies and resea rch communities. The met hod of defining
network res ources within ForCES nodes is more effective and
comprehensive. Google, Ericsson, Huawei and so on, all
proposed the research issues of using IETF ForCES technology in
SDN [4, 9, 10].
IETF ForCES working group was established in 2001. The
technology is relatively mature. ForCES itself is a technology
limited within the network device node. However, it also can be
applied to the entire network to achieve the goal of SDN. The
architecture of ForCES-based SDN is shown in Fig. 1.
ForCES and OpenFlow are often introduced as two well-known
supporting technologies for SDN architecture [11, 12]. Both of
them follow the basic SDN principle of separation between the
control and data planes; and both standardise information
exchange between planes. The main differences are the protocol
interface and the forwarding model in the forwarding devices
(SDN infrastructure layer shown in Fig. 1). The detailed compare
can be found in [11, 13].
In ForCES-based SDN, controller interacts with FFNs (ForCES
Forwarding Nodes) by using ForCES protocol through C-FFN
communication channel. This channel plays an important role in
ForCES-based SDN architecture. If Controller or FFNs cannot
respond the messages sent by each other in time, the system’s
efficiency and reliability will decline. This will directly affect the
system’s availability, and when the situation deteriorated, the
system even may be crashed.
The performance of C-FFN channel is mainly affected by the
SCTP congestion control mechanism of TML (Transportation
Mapping Layer) and ForCES message scheduling strategy. For a
specific application in ForCES-based SDN, the SCTP congestion
control parameters cannot be changed dynamically. The author of
this paper had created a scheduling and performance model for the
communication channel in ForCES system [14, 15]. Based on the
pre-research results, this paper is focused on the performance
evaluation, influence factors analysis, and performance
optimisation to the C-FFN channel.
The current researches on ForCES and SDN generally focus on
the implementation. Fewer studies are related to the performance
analysis. The research team in Zhejiang Gongshang University has
done some works on the ForCES performance. Zhou et al. studied
the problem of routing optimisation in ForCES router based on the
traffic matrix [16]. Li et al. studied a reliable multicast for
transmitting the ForCES messages, and verified its performance
based on the petri-net model [17]. Sheng et al. used the petri-net
to analyse the ForCES protocol [18]. Tootoonchian et al. showed
that simple modifications to the NOX controller (an
openflow-based SDN controller) boost its performance by an order
of magnitude on a single core [19]. Azodolmolky et al. selected a
hierarchical approach for the performance evaluation study to
address the scalability issue of SDN deployment [20]. These
above mentioned studies are all irrelevant to the communication
channel between the controller and the forwarding nodes. The
main contributions of this paper are: a performance model of
IET Networks
Research Article
IET Netw., 2016, Vol. 5, Iss. 6, pp. 162–169
162
&
The Institution of Engineering and Technology 2016