Available online at www.sciencedirect.com ScienceDirect Procedia Computer Science 155 (2019) 456–461 The 14th International Conference on Future Networks and Communications (FNC) August 19-21, 2019, Halifax, Canada Automation of network simulation: concepts related to IPv4 and IPv6 convergence Ayoub BAHNASSEa,∗, Faycal BENSALAHb , Fatima Ezzahraa LOUHABa , Azeddine KHIATc , Yousaf KHIATd , Mohamed TALEAa a Lab LTI, FS Ben M’sik, University Hassan II of Casablanca, Morocco STIC, Faculty of Sciences, Chouaib Doukkali University, El Jadida, Morocco c Lab SSDIA, ENSET Mohamedia, University Hassan II of Casablanca, Morocco d Center for Doctoral Studies ”Man-Society-Education”, Faculty of Education, Mohamed V Souissi University, Rabat, Morocco b Lab. Abstract The network simulation is a technique to test the operation of a technology or protocol before its actual deployment, or even during deployment. Several advantages can be derived from the network simulation, such as the saving of time and budgets. However, the simulation can sometimes represent certain limits if the concept to be simulated is large-scale or very advanced whose simple setting wrong can question the reliability of the results. These concepts include the coexistence of the two IPv4 and IPv6 protocols in a large-scale network such as the internet. No one can deny that most Internet Service Providers (ISP) provide only an IPv4based access network, however, client sites can use IPv6 locally. Several transition mechanisms and address translation techniques can be used to successfully coexist two different protocols. In this paper, we propose a new approach to automate the simulation of IPv6 and IPv4 transition mechanisms in a large scale network. The proposed approach is compatible with the Riverbed Modeler simulator and it may be modified to include other network simulators. The results of the evaluation of the approach show that the duration of setting up a network of 100 interconnected IPv6 nodes over an IPv4 network does not exceed 30 seconds in total. © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Conference Program Chairs. Keywords: Simulation; simulation automation; IPv6; Riverbed Modeler; tunnels; 1. Introduction IT tools are evolving rapidly and significantly, and their roles are crucial in almost every industry. These tools make it possible to visualize, calculate, predict, and improve the performances of a given system. Simulation tools are one of the main elements of the three phases of a project (before, during, and after). These allow to justify some technical ∗ Corresponding author. Tel.: +0-000-000-0000 ; fax: +0-000-000-0000. E-mail address: a.bahnasse@gmail.com 1877-0509 © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Conference Program Chairs. 10.1016/j.procs.2019.08.063 Ayoub Bahnasse et al. / Procedia Computer Science 155 (2019) 456–461 457 choices and also their impact on the performance of the entire system. A simulation tool can also be used to predict the behavior of a system under different conditions. For that, it will be necessary to model the system according to the real one and to master well its entries in order to have reliable results. The complexity of the simulation can be summarized, among other things, in the form of four points: 1. 2. 3. 4. Mastery of the simulation tool, its environment, language, and options. Mastery of the concept to be simulated and its application on a given simulator. The difficulty of setting up a large-scale network consisting of hundreds or thousands of equipment for example. Unsecured confidence interval, a simple misconfiguration can compromise all the results obtained. Despite the complexity of the simulation, the latter remains an inevitable step when it comes to a concept related to the production of a company. Among the concepts that are difficult to simulate taking into account (i) the high number of equipment that it can contain, (ii) the number of heterogeneous technologies that it can use, and (iii) the number of parameters that it can support we cite the internet network or the campus network. The Internet is defined as a set of interconnected equipment for the purpose of conveying information between a source and a destination. The first protocol used in this network is IPv4, which no longer responds to the constant need for additional address ranges, security and, Quality of Service (QoS), hence the appearance of the new IPv6 protocol. IPv6 addresses the limitations of IPv4 but is not widely used by most Internet Service Providers (ISP). As a result, the intermediate network of the operator is sometimes found to provide only IPv4 only, however clients use IPv6. Fig. 1 [3] illustrates an example of the prefixes exchanged within the autonomous system to which we currently belong. Fig. 1: The number of IPv4 vs IPv6 prefixes generated for AS6713 autonomous system. The results obtained illustrate that no IPv6 prefix is exchanged within our current autonomous system. While to interconnect two IPv6 sites across a fully IPv4 infrastructure two methods can be used, either encapsulating an IPv6 packet into another IPv4 and sending the set through IPv4 (Tunneling) or translating an IPv6 address to an IPv4 address in the transmitting interface and vice versa at the receiving interface (Header Translation). We will briefly present some of the most used tunneling techniques: 1. IPv6 Manual: also called static tunnel is a point-to-point tunnel, used to allow IPv6-IPv6 communication by encapsulation in IPv4 [6]. 2. 6to4 : similar to the first technique, however, this method makes it possible to establish point-to-multipoint tunnels in a fully dynamic manner [7]. 3. IPv6 over GRE : the Generic Routing Encapsulation (GRE) tunnel is designed to encapsulate all type of OSI layer 3 traffic. In this technique, GRE encapsulates endpoint IPv6 addresses in public IPv4 packets. This tunnel is also used for stable links whose address plan is fixed, moreover, this tunnel exists only between two routers which makes this solution non-scalable [4]. 458 Ayoub Bahnasse et al. / Procedia Computer Science 155 (2019) 456–461 4. Broker : a tunnel server is used to connect to the IPv6 network a dual stack machine located in an IPv4 network. To do this, you usually have to register with the tunnel broker service and then ask to open the tunnel. Then the tunnel broker will configure one of its routers to set up the tunnel [5]. The rest of the paper is organized as follows: in section two, we will describe the problem and position our contribution. In section three, we will present the related works. Section four will be dedicated to the presentation of our contribution. In section five we will evaluate the performance of our solution, and we will conclude in section six. 2. Positioning of the contribution The simulation of a large-scale network of two IPv4 and IPv6 protocols is a relatively complicated task because it requires addressing skills, routing, and the application of best practices for each of these protocols. Several factors make the simulation difficult, namely the size of the end network and the intermediate network, the IPv4 and IPv6 routing protocols used, and the tunneling mechanisms to be implemented. In addition, simulation requires certain skills: 1. The administrator must set up equipment, interconnect and develop two IPv4 and IPv6 addressing plans. 2. The administrator should be familiar with the steps for deploying IPv4-IPv6 transition techniques in Riverbed modeler. 3. The administrator must repeat the same scenario multiple times with different transition methods. 4. The administrator must be careful with the parameters of the simulation so as not to compromise the reliability of the simulation results. Our contribution consists in proposing a new approach for automating the simulation of IPv4 IPv6 tunneling mechanisms in a large-scale network. 3. Related work Several research projects have been carried out proposing platforms for automating the simulation of different technologies and network concepts. However, according to our research, no work has been done automating the simulation of IPv6 transition mechanisms. Bahnasse et al. [1] propose a server-based architecture for the simulation of virtual private networks. The proposed architecture collects user data and converts it into an XML file compatible with the OPNET Modeler simulator. The same concept was applied by Joze Mohorko et al. [8] for the automation of tactical network simulation under the same simulator. Recently the field of simulation automation has attracted the attention of a broad community of researchers. Hafsa et al. [9] proposed an architecture automating the simulation of heterogeneous and homogeneous wireless networks under OPNET Modeler, the architecture supports a variety of technologies namely WiMax, Wi-Fi, and Ethernet. The work [10] has been performed proposing the automation of the generation of network scenarios compatible with the OMNET ++ simulator. Bahnasse et al. [2] proposed an architecture based on automating the simulation of MPLS VPN networks by varying a set of parameters such as the size of the network, the type of architecture, and the applications used. 4. Proposed architecture The proposed architecture is based on three logical layers: the data plane, the plane control, and the management plane. Fig. 2 illustrates all the planes of architecture. Ayoub Bahnasse et al. / Procedia Computer Science 155 (2019) 456–461 459 Fig. 2: Architecture of the IPv4 / IPv6 transition mechanisms simulation automation solution. • Topology definition : this module makes it possible to define the network to simulate. It admits several parameters described in the management plane, among which we mention the number of bridges constituting the intermediate network as well as the number of bridges representing the gateways located at the customer sites. The interconnection of the various equipment of the intermediate network can be carried out according to predefined profiles by default (completely meshed, partially meshed or in a ring) or according to a personalized choice by the user. The interconnection can be carried out according to links of different categories (10/100 / 1000baseT, E1, T1, DS3, OC3, etc.). This module also defines the clients and servers used in the architecture as well as the simulated applications. Other parameters can be provided through this module such as the size of the network zone and the equipment manufacturers. • Addressing and routing : this module defines the version of the IP protocol to be applied for each network domain (intermediate and end). A calculation of the subnetworks is carried out for each segment, then an autoassignment of addresses is applied according to the elaborate addressing plan. The user can optionally specify the network addresses to be cut for each domain or segment. The choice of the routing protocol is then carried out according to the same logic of application of addresses. We note that our architecture supports the following protocols: Static, RIP, OSPF, EIGRP and IS-IS. • Tunneling method : this module allows to choose the tunneling method to apply between a source client site and a destination. Several tunneling methods are supported by our architecture, including those supported by the Riverbed Modeler simulator, such as a manual tunnel, 6to4, GRE, and Broker. For the first three methods, a tunnel interface is created, an addressing plan is then developed and assigned according to the communicating sites. • Statistic parameters : this module allows you to specify the statistics to collect. The statistics can be global (the general behavior of the entire network), per node (the performance of a specific device in a network), or by link (the performance of a link, the delay in the queue, bit error per packet, utilization, debit). • Recurrence : this module makes it possible to generate a set of scenarios by varying one or more parameters. This module can be useful if the user wants to test the scalability by increasing the number of nodes or the efficiency of one method compared to the other by keeping the same network architecture. • XML translator : this module makes it possible to translate the preceding parameters into an XML file compatible with the Riverbed Modeler simulator. 460 Ayoub Bahnasse et al. / Procedia Computer Science 155 (2019) 456–461 5. Performance evaluation In this section, we will evaluate the reliability of our approach in terms of speed of generation of several different scenarios. The first evaluation is to increase the number of nodes (intermediate and end routers) and the second is to vary the transition methods for each scenario. According to our research, no solution has been made automating the simulation of IPv6 transition mechanisms, we compared the efficiency of our approach compared to the manual method. Fig. 3 illustrates the delay in setting up a project consisting of five scenarios whose number of nodes is incremented by 20. Fig. 3: The duration of setting up a project by incrementing the number of nodes. Fig. 4: The duration of setting up a project by varying tunneling methods. It is clear that our approach has much shorter implementation times than the manual method. For the case of increasing number of nodes keeping the same transition mechanism IPv6 GRE. For the scenario consisting of 100 nodes, our approach requires 56 seconds while via the manual method a delay of 168 minutes is necessary. The gradual increase in the delay of our approach is justified mainly by the number of times the application looks for the appropriate XML tags for a parameter. Regarding the results obtained in Fig. 4, we find that our approach can implement the different IPv6 IPv4 transition methods in a 100 node scenario and this in less than 90 seconds, while via the manual method this requires 280 minutes. Ayoub Bahnasse et al. / Procedia Computer Science 155 (2019) 456–461 461 6. Conclusion The IPv6 IPv4 transition mechanisms allow the coexistence between the old Internet which is always valid (IPv4) and the Internet future whose passage is inevitable (IPv6). Evaluating the performance of these transition mechanisms is therefore important before the actual implementation phase. Indeed, simulation is one of the solutions for this objective evaluation. However, simulation has several limitations, the latter requires expertise in terms of the simulator and the technology to simulate. This paper proposes a new approach for automating the simulation of the different transition mechanisms under Riverbed Modeler. The results obtained showed the effectiveness of our approach compared to the manual method in terms of speed of generation of projects ready to be simulated. In order to create a 100-equipment project consisting of 4 scenarios each of which has a specific transition mechanism, the manual method requires 280 minutes while our approach reduces this time to 86 seconds only. References [1] BAHNASSE, A., EL KAMOUN, N., 2014. Policy-based automation of dynamique and multipoint virtual private network simulation on opnet modeler. Policy 5. [2] Bahnasse, A., Talea, M., Badri, A., Louhab, F.E., 2018. New smart platform for automating mpls virtual private network simulation, in: 2018 International Conference on Advanced Communication Technologies and Networking (CommNet), IEEE. pp. 1–8. [3] bgp.he.net, I., . AS36903 Office National des Postes et Telecommunications ONPT (Maroc Telecom). . [4] Conta, A., Deering, S., 1998. Generic packet tunneling in IPv6 specification. Technical Report. [5] Durand, A., Fasano, P., Lento, D., 2001. Ipv6 tunnel broker [rfc 3053]. [6] Gilligan, R., Nordmark, E., 2000. Transition mechanisms for IPv6 hosts and routers. Technical Report. [7] Kaeo, M., Rey, E., 2018. Opsec e. vyncke, ed. internet-draft cisco intended status: Informational k. chittimaneni expires: April 25, 2019 wework . [8] Mohorko, J., Klampfer, S., Fras, M., Cucej, Z., 2011. Expert system for automatic analysis of results of network simulation, in: Expert Systems for Human, Materials and Automation. IntechOpen. [9] Oulahyane, H.A., Bahnasse, A., Talea, M., Louhab, F.E., Al Harbi, A., 2017. Simulation automation of wireless network on opnet modeler, in: First International Conference on Real Time Intelligent Systems, Springer. pp. 237–249. [10] Virdis, A., Vallati, C., Nardini, G., 2016. Automating large-scale simulation and data analysis with omnet++: Lession learned and future perspectives. arXiv preprint arXiv:1609.04603 .