Professional Documents
Culture Documents
net/publication/319162733
CITATIONS READS
10 178
3 authors:
Ayoub Bahnasse
University of Hassan II of Casablanca
82 PUBLICATIONS 277 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Novel Frameworks of Smart and Adaptive QOS Management for computer networks View project
All content following this page was uploaded by Ayoub Bahnasse on 26 August 2017.
Abstract
This paper studies and evaluate Scalability of Voice over IP (VOIP) on various MPLS Tunneling technologies. Several works
have been carried out to evaluate the performances of VOIP in different IP and MPLS architectures. According to our re-
search, no scientific work has been done analyzing scalability by increasing the load of VOIP packets in different MPLS
tunneling technologies. This was a motivation for us to propose a scalability study of the VOIP technology in the various
scenarios (IP, MPLS, MPLS VPN and MPLS VPN protected by IPsec). The study was conducted on 64 different scenarios by
changing MPLS tunneling technology and rising the packet load to the order of 2 ^ n ranging from 64 to 475000. The ob-
tained results showed that the IP network is affected by a high latency and a bad MOS score. While MPLS remains scalable
even in high load scenario (475000 bytes). IPsec on MPLS VPN isn’t a good choice to add additional security level, from
results it’s shown clearly that from scenario of 46875 bytes on load, VOIP become unusable. This study carried out under
OPNET Modeler 14.5, the evaluation criteria are jitter, latency, loss rate and MOS score.
1. Introduction
Currently, Internet service providers are investing heav-
ily in modern telecommunications networks because of
their utility, ease of use and integration of new service
lines. This investment is mainly due to the strong trend
of emerging communications technologies such as VOIP,
Videophone etc... These technologies have become part
of every citizen’s daily life, and this is evolving over time.
According to the National Telecommunications Figure 1. Households’ Equipment with Computer by Area
Regulatory Agency (ANRT) “54.8% of households are of Habitation (% Households).
equipped with a computer/tablet. This proportion varies
according to the area (urban or rural), with nearly 7 in 10 Taking into account the explosive frequency of using
urban households versus one quarter of rural households Internet, so that infrastructures can follow these develop-
(Figure 1)”. ments, routing technologies must become fast, reliable
and support new services such as VOIP that are sensitive The study was carried out under OPNET Modeler, the
to several constraints such as: evaluation criteria are: jitter, latency, loss rate and MOS
• Jitter: The difference in delay between the sending of score.
two successive packets. The rest of the paper is organized as follows: in the
• Latency: The end-to-end delay between sending and second section, we will briefly study the various MPSL,
receiving the same packet. MPLS VPN and MPLS VPN protected by IPsec. In the
• Loss rate: The ratio between the number of packets third section we will present the state of the art and our
sent and received. motivations. The simulation environment will be detailed
• The MOS score: scale from 0 to 5 evaluating the qual- in the fourth section, describing the tool used, the scenar-
ity of the VOIP, 5 represents the perfect quality. ios chosen and the network parameters. The fifth section
Because of its reliability and quality of service, Multi- will be dedicated to the results obtained and we will con-
Protocol Label Switching (MPLS)1 technology takes clude in the last section.
advantage of other forwarding protocols in wide area net-
works. In an MPLS network applications traffic such as
voice and video follows a multidirectional model. Users
2. MPLS Technologies
can specify the minimum latency, jitter, and packet loss Initial Internet deployment responds to data transfer
thresholds for each type of traffic (voice, video, email, over the network. For this, a simple software router and
etc.), Subsequently, the MPLS network gives priority network interfaces were sufficient. As the ability to sup-
to the traffic of latency-sensitive data, such as voice and port high-speed transmissions has emerged, elements
video, compared to the rest of the traffic, which is less sen- capable of switching to Level 2 and Level 3 have had to be
sitive. deployed at the hardware level.
Unlike conventional tunneling technologies, data These solutions respond to the need for rapid packet
exchanges between the different sites do not transit via transfer while crossing the network, but do not meet the
the Internet but via the operator’s private network. MPLS service requirements for the information contained in the
VPN technology2 offers the security advantage by inter- packets. In addition, the majority of routing protocols
connecting multiple sites while guaranteeing QoS. In currently deployed rely on algorithms to obtain the fast-
addition, the implementation is very easy as the configu- est gateway on the network, but do not take into account
ration is done on the operator side. other measures such as delays or congestion that can
After discussing the constraints of VOIP and the seriously decrease Network performance. Traffic manage-
strengths of MPLS, the following questions should be ment is a goal for network administrators.
asked:
1. First, what is the difference between routing VOIP, in 2.1 MPLS
an IP network and an MPLS network;
A label3 identifies the path that the packet must follow.
2. What is the cost of the MPLS tunnel layer, ie the differ-
A label is transported or encapsulated in the level 2
ence between MPLS and MPLS-VPN;
header of the packet. The router that receives it, exam-
3. If the client wants to deploy IPsec, what will be the
ines the packet to determine the next hop according to
cost?
its label. Once a packet is labeled, the rest of its transfer
4. And most importantly, what is the degree of scalability
is based on label switching. The label values have local
of the VOIP in the various scenarios cited above.
significance. These values can also directly determine
Several research work has been carried out conduct- a virtual path (DLCI in Frame Relay or VCI and VPI
ing comparative studies between MPLS, MPLS VPN and in ATM4).
IP technologies. But according to our research none of The labels are associated with a FEC according to
them took into consideration the rise in charge in terms a logic or a policy determining this association. This
of packet size. This is for us a motivation to complete the decision can be made on the following criteria: Unicast
previous work in order to know who is most resistant to routing to destination, traffic management, multicast,
the load rise and when the VOIP is becomes unusable for Virtual Private Network (VPN) or QoS.
each scenario. MPLS performs the following functions:
2 Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology
Faycal Bensalah, Najib El Kamoun and Ayoub Bahnasse
• It is independent of the layers 2 and 3 protocols. Table 1 represents a comparative synthesis between
• It interacts with existing routing protocols, such as VPN technologies
RSVP5 and OSPF.
• Supports layer 2 layer of IP, ATM, and Frame Relay Table 1. Comparative synthesis between VPN technologies
networks. VPN MPLS IPsec VPN
In MPLS, data transmission is done on label-switched Provider Network Private Public
paths (LSPs). LSPs are a sequence of labels at each node Latency Low delay Non-Guaranteed
of the path from the source to the destination. LSPs are Throughput 100% 15% is consumed by IPsec
established according to the type of data transmission QoS Guaranteed very difficult (the data is
(control-driven6) or after the detection of a certain type of encrypted)
data-driven7. Labels are distributed according to the LDP Management Yes No
(Label Distribution Protocol), RSVP protocol, or by rout-
ing protocols such as BGP or OSPF. 2.4 Different Types of MPLS VPNs
That there are two types of VPN MPLS-based services:
2.2 MPLS Architecture Layer 3 MPLS VPNs and Layer 2 MPLS VPNs. Layer 3
MPLS has two major components: MPLS VPNs operate at Layer 3, and Layer 2 MPLS VPNs
• Control plane: Exchanges Layer 3 routing information operate at Layer 2 of the OSI model.
and labels; contains complex mechanisms to exchange
routing information, such as OSPF, EIGRP, IS-IS, and
2.4.1 Layer 3 MPLS VPNs
BGP, and to exchange labels; such as LDP, and RSVP
• Data plane: Forwards packets based on labels; has a Service providers typically refer to Layer 3 (L3) MPLS
simple forwarding engine VPNs when they say “MPLS VPN.” These VPNs are
MPLS is already used in many different applications: popular because they are the most scalable service pro-
• Unicast IP routing
vider option. In this scenario, your customer edge (CE)
• Multicast IP routing
routers exchange routes with your provider edge (PE)
• MPLS TE routers. When you use L3 MPLS VPN services, the
• QoS
service provider routers form the core of your WAN
• MPLS L2/L3 VPNs backbone. The MPLS VPN backbone always uses Border
Regardless of the application, the functionality is Gateway Protocol (BGP) as its routing protocol. Almost
always split in to the control plane and the data (forward- any other routing protocol can be used to connect your
ing) plane: sites with the MPLS VPN backbone, but many service
• The applications differ only in the control plane. providers limit the choices to BGP and static routing.
• The applications all use a common label-switching Each of your edge routers peers with just one router
data (forwarding) plane. -- the PE-router -- and you get optimum any-to-any
• Edge LSR Layer 3 data planes may differ.
connectivity between your sites regardless of your net-
work topology.
2.3 MPLS VPN
The MPLS VPN is a Private VPN that does not come out 2.4.2 Layer 2 MPLS VPNs
on the Internet (hence better security). It is mastered by a Layer 2 (L2) MPLS VPNs resemble a virtual circuit type
single operator and traverses a limited number of servers service and are very effectively used by service providers
controlled by the operator himself. This mode of trans- in the Metro Ethernet field. According to Enders, there
port has many advantages; it allows among others to give are two main RFCs that define two L2 MPLS VPN topolo-
priorities on flows (QoS plan: quality of service). gies:
IPsec VPN also allows the creation of a private network • The Martini draft specifies the concept of virtual cir-
between two or more sites, but the data is routed via the cuits as another overlay label switched path (LSP)
Internet, which means that the data can pass through sev- inside a tunnel LSP. It addresses the problem of point-
eral operators through a non-fixed number of gateways. to-point VPN connections in MPLS VPNs.
Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology 3
Scalability Evaluation of VOIP over Various MPLS Tunneling under OPNET Modeler
• The second important RFC on the label distribution VPN. The study will be done under the OPNET Modeler
protocol (LDP), drafted by Marc Lasserre and Vac tool.
Kompella, specifies the virtual private LAN service Indeed there are several scientific simulators that can
(VPLS), which presents a solution for multipoint con- be used as NS3, BONeS, etc. The author Jing-bo15 makes
nectivity for the Layer 2 MPLS VPN. It builds upon an objective comparison between them. From a database
the Martini approach by expanding the concept to a point of view, OPNET supports the equipment of differ-
full mesh network topology. VPLS is commonly mar- ent manufacturers, and especially it remains the simplest
keted under the name “Enterprise Private LAN.” tool to handle.
4 Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology
Faycal Bensalah, Najib El Kamoun and Ayoub Bahnasse
Table 3. Equipment and technology of links used in 5.2 Loss Rate
simulation
P c7200
PE c7200
CE c3745
Link CE-PE FastEthernet
Link PE-P GigaEthernet
Link P-P GigaEthernet
5.3 Latency
Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology 5
Scalability Evaluation of VOIP over Various MPLS Tunneling under OPNET Modeler
Figure 5 illustrates the latency results. Taking into scenario of 2048 bytes on load the IP scenario bypassed
account the results of the loss rate, it can be seen that the the acceptable VOIP values, while MPLS remains scalable
IP protocol is not efficient even if its latency is the lower even in high load scenario (475000 bytes).
compared to the other scenarios. IPsec on MPLS VPN isn’t a good choice to add addi-
The MPLS protocol offers the best results, the same tional security level, from results it’s shown clearly that from
results are shown from MPLS VPN. The difference scenario of 46875 bytes on load VOIP become unusable.
between them is justified by the process of routing in the
tunnel and the dual labeling process performed in MPLS
VPN.
7. References
IPsec, as in the IP network, adds an additional delay, 1. Davie BS, Rekhter Y. MPLS: technology and applications.
making VOIP unusable from fourteenth scenario. Morgan Kaufmann Publishers Inc; 2000.
2. Chen W, Chen H, Liu KH, Soon SC, Zhou B. U.S. Patent
5.4 MOS No. 7,450,598. Washington, DC: U.S. Patent and Trademark
Office; 2008.
3. Wijnands IJ, Minei I, Kompella K, Thomas B. Label
Distribution Protocol Extensions for Point-to-Multipoint
and Multipoint-to-Multipoint Label Switched Paths (No.
RFC 6388). 2011.
4. Pham CD, Fdida S. Evaluation de Performance des Reseaux
ATM: Etude et Perspectives en Utilisant la Simulation
Distribuee. In: Actes du Colloque Francophone sur
l’Ingénierie des Protocoles (CFIP’96), 1996. p. 17–31.
(a) IP Scenario (b) MPLS Scenario 5. Zhang L, Deering S, Estrin D, Shenker S, Zappala D. RSVP:
A new resource reservation protocol. IEEE network. 1993;
7(5):8–18. Crossref
6. Das, Saurav, Sharafat, Reza A, Parulkar, Guru, et al.
MPLS with a simple OPEN control plane. Optical Fiber
Communication Conference and Exposition (OFC/
NFOEC), the National Fiber Optic Engineers Conference,
2011. p. 1–3. Crossref
7. Moisand, Jerome P, Aggarwal, Rahul, Wadhwa, Sanjay, et
al. Layer two (L2) network access node having data plane
(c) MPLS VPN scenario (d) MPLS VPN IPsec MPLS. U.S. Patent No 8,121,126, 21 févr. 2012.
Scenario 8. Ravonimanantsoa N, Rakotomalala M.A. Dimensionnement
Du Backbone Mpls Pour Les Traffic Voip. Madaeti. 2015;
Figure 6. MOS. 2:85–94.
9. Porwal MK, Yadav A, Charhate SV. Traffic Analysis of
Figure 6 shows the MOS score. MPLS offers the most MPLS and Non MPLS Network including MPLS Signaling
suitable score, while IP and IPsec based scenarios offer Protocols and Traffic distribution in OSPF and MPLS.
the worst quality of speech. MPLS VPN offers an accept- First International Conference on Emerging Trends in
able score. Engineering and Technology, ICETET’08, 2008. p. 187–92.
Crosserf
10. Sllame AM, Aljafari M. Performance Evaluation of
6. Conclusion Multimedia over IP/MPLS Networks. International Journal
of Computer Theory and Engineering. 2015; 7(4):283.
Under OPNET Modeler, in this study we evaluated the 11. Jannu K, Deekonda R. OPNET simulation of voice over
scalability of architectures: (i) IP, (ii) MPLS, (iii) MPLS MPLS with Considering Traffic Engineering. Blekinge
VPN, and (iv) MPLS IPsec VPN. Measurements were Institue of Technology. 2010; 15.
made by increasing the VOIP packet load. 12. Akinsipe O, Goodarzi F, Li M. Comparison of IP, MPLS and
The results obtained showed that the IP network is MPLS RSVP-TE Networks using OPNET. International
affected by a high latency and a bad MOS score. From Journal of Computer Applications. 2012; 58(2). Crossref
6 Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology
Faycal Bensalah, Najib El Kamoun and Ayoub Bahnasse
13. Faisal M, Uddin J, Shil S. Performance of VoIP networks 15. Xia JB, Li MH, Wan LJ. Research on MPLS VPN network-
using MPLS traffic engineering. Advanced Materials ing application based on OPNET. International Symposium
Research. 2012; 457:927–30. Crossref on Information Science and Engineering, ISISE’08. 2008;
14. Ali S, Rana BZ. OPNET analysis of VoIP over MPLS VPN 1:404–8.
with IP QoS. Sweden: Blekinge Institute of Technology;
2011.
Vol 10 (29) | August 2017 | www.indjst.org Indian Journal of Science and Technology 7