Session S2C USING COMPUTER NETWORK SIMULATION TOOLS AS SUPPLEMENTS TO COMPUTER NETWORK CURRICULUM Nizar Al-Holou1 , Ph.D., Knicolas K. Booth2 , and Ece Yaprak 3 , Ph.D.
Abstract - Computer network technologies have been growing explosively. Teaching computer networking principles can be enhanced using simulation through the use of interactive (not statistical based) simulation. The curriculum is based on the theory that through networking simulation programs, students are able to graphically implement the concepts learned in their coursework. With networking simulation tools, students can construct, tune, and analyze network performance while reinforcing their understanding of networking theory. Network simulation tools save money by offering the user a tool to enhance the design and integration of networks. There are many network simulation tools that allow users to model LANs, MANs, and WANs on the market. This paper describes our approach to develop computer network laboratory that enhances the understanding of a computer network course. Different computer network simulation tools have been considered and COMNET III has been selected as the tool of choice. Five labs have been developed. COMNET III is a network simulation tool that allows students and networking professionals to model, tune, and analyze the performance of various types of networks. It fosters what COMNET refers to as a “building-block approach” by utilizing a GUI interface that is comprised of “blocks” or objects representing many of the objects used in real networks. The user has the ability to adjust the objects’ parameters using either default, constant, or algebraic values to better model the objects’ functions
INTRODUCTION The study of computer networks can be challenging. Students have to learn ever-changing technical jargon; master and apply concepts covered in textbooks, lectures, and seminars; and achieve these objectives in a fifteen-week period. Many students complain about the length of time allowed to effectively cover material and are frustrated with the way some computer network textbooks are written. When given the opportunity, lab work provides the students with much needed time to absorb the material and learn how to implement it in the real world. However, it is costly to develop a computer laboratory. Simulation tools can be used to enhance students’ understanding and replace some aspects of lab.
There are many different network simulation tools such as COMNET III, OPNET, CNET, and Prophesy. When determining which tool to use for computer network lab, we focused on several necessary attributes; inexpensive, userfriendly, performance, and meshes well with our coursework. OPNET of Mil 3, Inc. provides a
comprehensive development environment supporting the modeling and performance evaluation of communication networks and distributed systems [1]. It provides extensive support for developing models of new components with programming language called Proto-C. The best feature that OPNET offers it the flexibility to develop detailed custom models since it comes with all its Proto-C code. This tool allows the user to simulate practically any system involving requests, services, and queues [2]. However, OPNET simulation
package is not easy to learn in one semester and we feel that it is more appropriate for graduate courses. There are high performance tools, such as NetMaker XA from Make Systems, Inc. and Optimal Performance from Optimal Networks. Both tools designed to simulate router, bridge, protocol, and application behavior. However, they differ from most traditional simulation software [3]. Traditional simulation networks require the user to manually enter and update network data. NetMaker XA and Optimal Performance use actual network protocol information retrieved by probes to automatically interpret and design the network. As an added feature, Optimal Performance issues specific recommendations for improving network performance. However, the expense related to software of this caliber is expensive for universities who aspire to incorporate network simulation tools into their coursework. Optimal Performance comes in two versions, Network and Enterprise [4]. Although these tools are highly accurate and offer many key features, for our purposes neither tool presented itself as a cost-effective educational network simulation tool. Our coursework does not necessitate a need for the probes that both applications boast contribute to their amazing accuracy in the collection of data. Many of our labs are centered on teaching the student how to build their own network using data they collect and construct themselves. Secondly, Optimal Performance’s ability to recommend network improvements was another feature that
1
Electrical and Computer Engineering, University of Detroit Mercy, Detroit, MI 48219,
[email protected] Electrical and Computer Engineering, University of Detroit Mercy, Detroit, MI 48219 3 Department of Engineering Technology, Wayne State University, Detroit, MI 48202 2
0-7803-6424-4/00/$10.00 © 2000 IEEE October 18 - 21, 2000 Kansas City, MO 30 th ASEE/IEEE Frontiers in Education Conference S2C-13
Session S2C clashed with our educational aims in this course. The exercises are structured to test the student’s understanding of performance issues within a network; therefore, Optimal Performance did not meet our needs as an educational tool. The labs and exercises are structured with the intent of getting students to analyze and evaluate network performance and, therefore, to decide for themselves what methods are best to improve networks. Furthermore, we preferred a less expensive application to use for our labs. CNET is a network simulation tool that allows experimentation with various OSI/ISO Networking Reference Model layers, such as the data-link, network, and transport layers. Developed by Dr. Chris McDonald, a professor at the University of Western Australia, CNET is supported by many UNIX and UNIX variants; however, a Windows-based version is still under development. CNET supports a model of networking in which nodes representing either workstations or routers are connected by point-topoint links not by shared physical media, such as buses or rings [5]. It requires network protocols to be written in either the K&R or ANSI-C programming language and supports their execution within a single UNIX process [6]. Although CNET fostered many of the attributes we were looking for, the complexity associated with a command line interface when compared to a GUI interface led us to consider future work with this tool; but our current work compelled us to search for a simpler program. COMNET III provides an integrated, graphical environment for modeling and performance analysis of computer networks [7]. It allows the user to interact with the model during execution. Students are able to graphically display traffic modeling patterns, use realistic network objects to reflect real networks, and apply the concepts discussed in lecture. COMNET III contains a standard Windows ™ interface for building network models. Its GUI interface includes a palette that comprises all of the objects used for manufacturing network models, a standard menu format from which you may activate pull down menus that encompasses most of the windowed applications [8]. Its menus are fairly simple to navigate through, and COMNET III includes a complete and thorough reference manual as well as online technical support.
COMNET III COMNET III provides detailed modeling of network object logic. An object’s I/O subsystems, databases, and the applications that run on the systems can be simulated using this tool. Through the usage of the user’s description of the network, its routing algorithms, and workload, COMNET III simu lates the operation of the network and provides measurements of network and modular performance. COMNET III was developed using the programming language MODSIM II with an object-oriented design. It was designed with several objects whose characteristics could be modified to match those of real network objects. With
COMNET III you can create a variety of network architectures, including LANs, WANs, packet switching, ATM, frame relay, and so on. Moreover, you can define different network topology, such as Ethernet, Token Ring, and Token Bus [9]. Transport layer protocols such as connection-oriented and connectionless-oriented protocols can be selected. The user can segregate reports by object or layer and can select the reports need to be examined. In an effort to provide accurate and realistic simulation results, COMNET III uses discrete event simulation methodology. Analytical methods that develop mathematical based model cannot handle the effects of random variance. Moreover, the model should be approximated to fit the mathematical derivation. Such ability is paramount for the usage of the generated routing algorithms and tables used to model traffic on the network and between objects. In this paper, we will outline our approach to teaching computer networking using COMNET III as the simulation tool.
LABORATORY DEVELOPMENT The laboratory exercises have been developed to accomplish several objectives. The first objective has been to train students how to navigate through different menus and windows of this tool. The second objective has been to help students visualize how network concepts are applied in the real world by using graphical examples in conjunction with those used in the textbook. The third objective has been to allow students to see the data movement along data paths by tracing messages from the source to the router to its destination. COMNET III gives the user the option to animate simulations. The last objective has been to challenge the students to utilize their decision-making ability and rationalization to evaluate performance improvement options that are key to network administration. After each lab, students are required to run simulations monitoring the performance of the network and its components. Using the reports from the simulation, students are required to go through and find if there is any network bottleneck or performance problem that might occur. The students are expected to decide if the result shows an efficient network. If there is a performance problem, the students must then find ways to improve the model using the reports, as a guide to produce desired results. In the lab, students learn how to perform several tasks centered on network administration and engineering. They construct LANs, WANs, and internetworking systems. They test various static, adaptive, and user-defined routing algorithms to see the effect each algorithm has on traffic in the network model. By adding new applications to a network and isolating their testing to network message source/response nodes as well as the data link layer, students are able to evaluate the effect of introducing new applications has on the network as well as experience network sizing at the design level.
0-7803-6424-4/00/$10.00 © 2000 IEEE October 18 - 21, 2000 Kansas City, MO 30 th ASEE/IEEE Frontiers in Education Conference S2C-14
Session S2C The labs have been intertwined with one another, using sometime previously designed models for future labs. This way we are able to further develop a concept by using older labs as a springboard for future analysis. Students may use the reports generated in the previous model to evaluate the changes that are currently taking place in the network and compare them with the original model’s results. This allows the student to further develop their performance evaluation skills by comparing their designs against the original model. In the first lab, “Familiarity with COMNET III,” students learn how to navigate through the application using the toolbar, pull down menus, and multiple windows in order to formulate the characteristics and parameters of the network models. Students use the toolbar to create the objects for the network mo del with the pull-down menus and windows they are able to define the parameters for each object and the network backbone. Some of the objects features in COMNET can be used for several kinds of objects for future reference; students are given working definitions for the objects used in the model. After the network model is complete, the student verifies the model to ensure correctness and runs the replication to test the model. The student selects various reports used for analyzing and evaluating the performance of the network and each of the nodes used in building the network. Through the first lab, students use default parameters to model traffic on the network. The second lab, “Creating User-Defined Parameter Sets,” entails having the student generate two networks using user-defined parameters to model traffic on the network. The student uses the network generated in Lab 1 and adjusts the parameters according to Lab 2’s new user-defined parameters. After completion, the new network models are tested; the student then uses the reports from Lab 2’s network model and the network constructed in Lab 1 to compare and contrast each network’s strengths and weaknesses. Lab 3, “Using Routing Algorithms to Model Traffic on a Network,” simulates a packet switching network using various routing algorithms. The student builds three networks, each of which utilizes different routing algorithms. Students then test the networks for delays, collisions, and terminated packets to determine which network is the most efficient. Lab 4, “Adding New Applications on a Network,” uses one of the networks created in Lab 3, adds a LAN Voice Mail application to the model, and tests the new model to determine what effect the new application has on network traffic. In Lab 5, “Wide Area Network Simulation using COMNET III,” students build two WANs; each has three subnetworks that will utilize desktop video teleconferencing using MBONE (Multicast Backbone). The students will test the effectiveness of using frame relay as a packet switching method of this kind of network. The students utilize the reports to evaluate the performance of the proposed network. COMNET III
generates two kinds of reports, Trace and Statistic. The Trace report records all of the packet exchanges between objects and includes the time of the initial packet a well as final acknowledgement of receipt. Using the Trace report, the student checks for any collisions that occurred during the simulation period. The Statistic reports collect the statistics for the reports you selected for analysis. It records how many packets were generated at each node or object; the amount of traffic experienced for various topologies, collisions, and retransmittals; how many messages or responses were generated by node; and other data per request. By using the reports generated in the labs, the student investigates possible faults in the network. The student is able to formulate an opinion based on data in these reports that indicate the effectiveness of the network understudy. Students are then asked either to supply a better-constructed network based on the faults found and recorded by the student or prove its effectiveness using evidence provided in the reports. The student, therefore, applies reinforced concepts and reasoning to a network model in order to evaluate performance improvement options. Each lab has a few questions based on the reports generated from the simulation to test the students’ understanding as well as analytical and reasoning skills. In the exercises, students are also required to design their own network and test its performance against the models they built in the lab. This gives the student an opportunity to test and apply their knowledge of network concepts for the purpose of reinforcing the topics covered in the lab. They again are able to scrutinize the lab reports and gather information on what made it effective or what factors may have contributed to its ineffectiveness. The students determine what is necessary to better the network and attempt to build a more efficient network model using the criteria given.
CONCLUSIONS We have used COMNET III for the last two years in our lab. The students like its user interface and ease of use. Also, CACI provides an academic and student version at a very low cost, which makes it affordable to many institutions and students. The tool enables students to design, evaluate, and test different network architectures. Moreover, it reinforces network concepts presented in the classroom and the textbook that. Students are able to apply the concepts learned during their coursework to construct real network models for evaluation and research. COMNET III’s many features and utilities provide users with the necessary tools to examine networks by layer, from the physical layer to the application layer. Its usefulness as a network simulation tool has found success in many programs and in commercial markets as a network
0-7803-6424-4/00/$10.00 © 2000 IEEE October 18 - 21, 2000 Kansas City, MO 30 th ASEE/IEEE Frontiers in Education Conference S2C-15
Session S2C simulation tool used internationally by many medium-tolarge sized companies and universities. We have been working to include other simulation tools to provide students with other means of conducting network research, such as CNET. By adding this tool, not only are we able to provide students with another tool for simulating networks, we are also able to teach students how to use UNIX and UNIX variants, therefore expanding their wealth of computer knowledge and experience.
REFERENCES [1].
MIL 3. OPNET Modular Web Site. www.mil3.com/products/modular [2]. Abstra ction Software Web Site for Prophesy, www.sni.net/ [3]. Graziano, Claudia. “Simulation Tools Get Real.” LAN Times, 23 Jan. 1995. [4]. Streeter, April “Optimization for the Elite” LAN Times, 27 Mar. 1995. [5]. University of Ric hmond. Dept. of Computer Science Web Site CS395, www.mathcs.urich.edu/ ~barnett/cs395/cnet1.3/. [6]. Chris McDonald’s CNET Web Site, www.chris.uwa.cs.edu. [7]. Jones, Jeffry, “COMNET III: Object-Oriented Network Perfo rmance Prediction,” Winter Simulation Conference Proceedings, WSC’95, 1995. [8]. Sullivan, Jeffrey E. COMNET III™ Tutorial A Detailed Guide to Modeling Networks with COMNET III. Diss. Naval Postgraduate School, 1993. [9]. Cheung, Raymond, et. al. “Modeling and Simulation using network software to evaluate SCADA system Configuration,” IEE Proceedings of the 1997 4 th International Conference on Advances in Power Systems Control, 1997, pp. 609-613. [10]. COMNET III, University Edition Release 1.4, Computer Software Reference Manual, CACI, 1997.
0-7803-6424-4/00/$10.00 © 2000 IEEE October 18 - 21, 2000 Kansas City, MO 30 th ASEE/IEEE Frontiers in Education Conference S2C-16
