Professional Documents
Culture Documents
OSPF Cost Impact Analysis On SDN Network: Ronald Adrian
OSPF Cost Impact Analysis On SDN Network: Ronald Adrian
OSPF Cost Impact Analysis On SDN Network: Ronald Adrian
Abstract— Software Defined-Network is a new technology in the services such as SNHX-IP for RYU controller [4]. There is a
network engineering. This technology allows a server which is called fundamental difference because the use of real network devices
by a controller and it controls all connected devices. All as part of the role of forwarding plane will get more realistic
configurations and resources of network devices become centralized results with conditions on the field. Through this topic is
to the controller. One of them is routing configuration. This makes expected to be produced output with equivalent parameters that
it easier for network administrators to configure routing on complex can determine the best choice in determining the routing
networks. This study focused on OSPF implementation and QoS protocol used in the network.
performance analysis on SDN networks. OSPF implementation can
be configured using cost and no cost. This configuration can be SDN is a new paradigm in network development [5].
done as a configuration of cost settings on conventional networks. It Research field of routing is done to analyze how the
affects the selection of main data paths on OSPF routing. This performance of routing protocols previously implemented in
implementation in this research used the Mikrotik devices. Data conventional networks and then implemented in new paradigm
retrieval involves convergence time and some QoS parameters such technology. SDN research is generally still conduct on mininet.
as Throughput, PLR, Jitter and Delay. In testing phase used traffic Mininet hosts run standard linux networking software and
data which generated by iperf and D-ITG with variations of existing switches that support OpenFlow for highly flexibility custom
data types. The goal of this research is to find the best routing routing and SDN.
configuration on SDN networks, especially in OSPF. This research
can be expanded with the various parameters and complex topology. Several studies have been done on the SDN network such as
SDN controller performance testing, POX and Floodlight.
Keywords : SDN, QoS, Delay, Jitter, Cost. Floodlight give better management of large amounts of data and
requires a very high data flow settings than the others. POX
provides a more constant handling guarantee for whatever
I. INTRODUCTION
number of hosts [6]. That research becomes one example of
In recent years SDN technology has become an interesting SDN research without routing function. On a large network,
topic for researchers. SDN technology is a network technology routing needs can not be avoided either on conventional
in which the infrastructure part of the device, ie control plane networks or SDNs. Routing requirements on the SDN network
and data plane is done separation, so the routing policy can be can be configured by some controllers, such as the use of
done centrally through the controller. Centralized network RouteFlow and SNHX-IP based on the RYU controller [7].
control SDN makes network settings easier and flexible. Current
SDN development has covered things such as load balancing, The use of RouteFlow controllers in research has been
VLANs, and one of them concerning routing protocols through successfully applied to several routing protocols, such as OSPF,
RouteFlow [1]. eBGP, and RIP. Research carried out through simulation routing
using the eBGP protocol and performed on the SDN network by
The architectural in SDN and conventional network will retrieving QoS (Quality of Service) data states meets ITU-T
make a fundamental difference from the performance sector. standards if it flows background traffic up to 75 Mbps. From the
This study was conducted by analyzing the performance of the test results also obtained some things, where the results of
existing OSPF routing protocols on SDN and conventional convergence time, and Routing overhead is influenced by the
networks. It based on cost and without cost by utilizing number of switches and features.
Routeflow [2]. OSPF routing protocol was selected because it is
one of the most widely used routing protocols. We use a variable This paper is organized as follows. In the next section, we
cost as a benchmarking instrument because OSPF is an LSA explain the previous and related work to our research. Section
routing protocol. The best route will be determined based on the III is the method. We provide result in section IV. Finally, we
cost value [3]. Similar research has also been done using conclude in Section V.
different protocol types, ie RIP and EBGP or using other
Figure 1. Non Cost OSPF Topology Figure 3. Data Delay Traffic on Non-Cost SDN
Based on Figure 3 above shows that the value of delay for
data service delivery is fluctuating, which has an average value
ranging from 0.006584 s - 0.009474 s for every multiple of
giving background traffic. However, the average on the graph
above is still in a stable range because both the decline and the
increase are not too significant. The above also shows that giving
background traffic on the link does not affect significantly.
Significant effect occurs when giving background traffic of 100
Mbps reaching 0.009474 s from the previous 0.006068 s when
background traffic is given at 90 Mbps. Overall average of data
Figure 2. Cost OSPF Topology service delay for scenario of SDN architecture usage without
modification cost 0,006725455 s.
Testing is done by generating various data types and
baground traffic using iperf and D-ITG. Data packet is captured
by wireshark. This research is implemented directly using
mikrotik devices. We analyze the delay and jitter from the above
topology.
Delay formulation can be shown as equation (1).
∆ & (1)
Explanation :
: one-way delay value
: transmit time
: receive time
Jitter formulation can be shown as equation (2), (3) and (4). Figure 4. VoIP Delay Traffic on Non-Cost SDN
Jitter is a delay variation which produced from various data
transmission. Other test results based on Figure 4 show that the delay value
for service delivery in the form of VoIP has increased with the
(2) giving of background traffic. The resulting average value ranges
(3) from 0.005917 s - 0.009271 s for any given traffic background
multiples. The trend of increase in delay value also tends to be
199
2017 2nd International Conferences on Information Technology, Information Systems and Electrical Engineering (ICITISEE)
stable, it is indicated starting from giving background of traffic traffic background, it shows an increase of 65%. This can occur
20 Mbps - 80 Mbps which range of value increase ranges from due to the influence of traffic load and the amount of congestion
3-19%. Significant effect occurs when giving background traffic in the network causing packet delay to vary. But the average
of 100 Mbps reaching 0.009271 s from the previous 0.006068 s value of data traffic jitter when the background traffic 40 Mbps
when background traffic is given at 90 Mbps or an increase of tend to experience a downward trend to 100 Mbps point. It is
23%. Overall average VoIP service delay for scenario of SDN noted that the decrease of jitter tedalam value when the
architecture usage without modification cost 0,007060909 s. background of traffic traffic of 100 Mbps, which became
0.000614 s from the previous 0.000764 s. This can happen
because the network has a bottleneck condition, so the received
packet has a relatively same arrival time.
200
2017 2nd International Conferences on Information Technology, Information Systems and Electrical Engineering (ICITISEE)
influence occurred when giving background traffic of 80 Mbps flown with 20 Mbps traffic background which causes a 35%
which happened decrease of 28% to 0,005194 s, after which increase. This happens because of the influence of traffic loads
happened rebound equal to 39% when flowing background 90 and large congestion within the network resulting in packet
Mbps. Overall average service delay of voip is 0,006564818 s. delay varies. After the background traffic of 30 Mbps the
average jitter value has decreased trend up to 100 Mbps. The
average percentage of jitter impairment trend occurred by 13%.
This can happen because the network has a bottleneck condition.
V. CONCLUSION
The application of configuration cost modification on the
SDN network has an influence on network performance.
Comparison of SDN and conventional network scenarios based
on two parameters (QoS and convergence time) using either cost
modification or not. It indicates that the best delay average value
is generated by the SDN network using the cost setting of 5.947
ms. The value is the lowest compared to others. Overall it shows
that SDN performs better than conventional when running OSPF
Figure 9. Video Delay Traffic on Cost SDN routing.
On the other side as shown in Figure 9 the resulting video
ACKNOWLEDGMENT
traffic delay values tend to experience an upward trend as the
background traffic flows. This is reinforced by the average This research is funded by funding capacity improvement of
percentage percentage increase of 7%. On average, sampling young lecturer of Universitas Gadjah Mada.
traffic flow of 20 Mbps load becomes the lowest delay value and
becomes the deepest percentage of decline of 26% before finally REFERENCES
rebounding on the flow of background traffic afterwards.
[1] G. N. Nde and R. Khondoker, “SDN testing and debugging tools: A
survey,” in 2016 5th International Conference on Informatics,
Electronics and Vision (ICIEV), 2016, pp. 631–635.
[2] A. Rego, S. Sendra, J. M. Jimenez, and J. Lloret, “OSPF routing protocol
performance in Software Defined Networks,” in 2017 Fourth
International Conference on Software Defined Systems (SDS), 2017, pp.
131–136.
[3] K. Németh, A. Kőrösi, and G. Rétvári, “Optimal OSPF traffic engineering
using legacy Equal Cost Multipath load balancing,” in 2013 IFIP
Networking Conference, 2013, pp. 1–9.
[4] R. K. Arbettu, R. Khondoker, K. Bayarou, and F. Weber, “Security
analysis of OpenDaylight, ONOS, Rosemary and Ryu SDN controllers,”
in 2016 17th International Telecommunications Network Strategy and
Planning Symposium (Networks), 2016, pp. 37–44.
[5] S. Chakraborty, S. Chakraborty, and S. Nandi, “Beyond conventional
routing protocols: Opportunistic path selection for IEEE 802.11s mesh
networks,” in 2013 IEEE 24th Annual International Symposium on
Personal, Indoor, and Mobile Radio Communications (PIMRC), 2013,
pp. 3224–3228.
Figure 10. Jitter Traffic on Non-Cost SDN [6] F. Yamei, L. Qing, and H. Qi, “Research and comparative analysis of
Figure 10 above shows the value of jitter derived from the performance test on SDN controller,” in 2016 First IEEE International
Conference on Computer Communication and the Internet (ICCCI),
sample data traffic. In the figure shows that the value of jitter 2016, pp. 207–210.
reaches the highest value when the measurement flows through [7] J. Stringer et al., “Cardigan: SDN distributed routing fabric going live at
the background traffic of 30 Mbps. It is characterized by the an Internet exchange,” in 2014 IEEE Symposium on Computers and
average sample value of jitter reaching 0.001543 s. However, Communications (ISCC), 2014, pp. 1–7.
the highest percentage increase occurs when the network is
201