Analyzing The Results

After running the simulation, you'll come across to viewing the results which is similar to the graphs shown below. The simulated network was implemented to compare the performance between well known routing protocols RIP, EIGRP and OSPF. As shown below, EIGRP out-perform all other protocols providing the fastest convergence duration with only 0.0136 second. The reason behind that is due to the effective mechanism of DUAL algorithm, which maintains the topology by exchanging small-size Hello packet rather than periodic updates. On the other hand, OSPF/RIP combination performed worst among the other protocols, as the convergence time in this case is the sum of OSPF and RIP networks. The figure depicts network convergence duration where X-axis represents convergence time in seconds and Y-axes represents simulation time in minutes. As opposite to expectation, OSPF network required 55 seconds to converge, this is probably because OSPF network have not been split into areas, which indicates that it is necessary to have a hierarchy design in order to get the most of OSPF.

Performance parameters

In this scenario, we will conducts a study of various performance metrics for each scenario. The scope of metric could be global or per-node basis. These metrics will be analysed to evaluate the performance of the various dynamic routing protocols. There are huge number of data that can be collected using OPNET, here's a list of the ones we will focus on.

·         Network reliability

•   Network Convergence Duration: to determine how much time is needed for all routers to have identical routing table. The less time the better.
•   CPU Utilisation: to determine which protocol require less CPU usage.
•   Delay: to determine which protocol provide less overall end-to-end delay.
•   Bandwidth Utilisation: to determine which protocol provide less overhead.
•   Traffic dropped: to observe how well each protocol deals with network failures
•   Link Throughput, Utilisation and Queuing delay: to determine which protocol perform best regarding link metrics.

·         Non-real time application

•  FTP Download Response Time: to measure the time elapsed between sending a request and receiving the response packet  Email Download Response Time: to measure the time elapsed between requesting an email and receiving it from the network’s Email server
•  HTTP Object Response Time: to calculate the response time for each object in HTML page.

·         Real time application

•   Video Packet Delay Variation: to measure the variance among end to end delays for video packets from the time it is created to the time it is received.
•   Video End-to-end Delay: to measure the accumulated time needed by all nodes to send a video application packet to a destination node’s application layer.
•   VoIP Jitter: by definition, jitter is the variation in the time between packets arriving. The study will show which protocols provides better jitter.
•  VoIP Packet Delay Variation: to measure the variance among end to end delays for video packets from the time it is created to the time it is received
•   VoIP End-to-end Delay: to measure the total voice packets delay in the network that includes network delay, encoding/decoding and compression/decompression delays.

Network Simulation configuration

We will simulate the network decribed in the previous post using OPNET. After adding all network devices, we should start configuring the nodes s follows:

Router configuration 

For every proposed scenario, each router will run only one routing protocol (e.g. RIP, IGRP…etc.) with the exception of protocol-combination scenarios, where routers will run two protocols simultaneously.

Application Configuration
Five applications will be configured to support FTP, Email, Web Services, Voice over IP (VoIP) and Video Conferencing. The services demand by workstation for FTP and Email is set to Medium Load while Web service is configured to be used as heavy browsing. FTP object size is set to 5 KB whereas it is 1 KB for Email. The HTTP page is configured to be 1 KB in size in additional to five medium size images per page. The image size is variable between 500 byte and 2KB. Voice quality is configured to meet Pulse Code Modulation (PCM Speech Quality) while Video application is set to Video Cassette Recording (VCR) with frame size of 352X240 pixels and 30 frames/sec qualities. All services are set to start at 100 seconds to evaluate them after network settlement. 

Network Simulation

We will start out first lesson in network simulation by creating a network map for UK consists of 32 sites. Each site will have one router, one switch, 2 loal area networks and 3 servers.

The simulated network will be geographically spread across 32 UK cities, each subnet consists of a router, a switch and one LAN comprises 25 workstations. Main subnets (London and Manchester)will contain servers to provide particular services such as FTP, HTTP and Email server. Additionally, for each main subnet, 10 workstations will be running VoIP and video conference services 

In the next lesson, we will learn how to configure the nodes such as switches, router and the servers as well as configuring the profile and application objects.

Data and statistics collection

OPNET Modeler provides a vast verity of statistics either globally or by mode basis. Global statistics are scoped to the simulation as a whole for the entire simulated network. For instance, to examine the Ethernet delay for the entire network, a Delay option from the global statistics is selected as shown in the figure.

In contrast, node statistics collect statistics on particular node (e.g. router). For instance, CPU Utilisation option will measure the CPU utilisation for a particular router. 

Node configuration

LAN workstation can run different profiles depending on the geographical area or services they entitled to use. For router to operate, a routing protocol must be assigned to it. Routers are capable of running most common IPv4 dynamic routing protocols such as RIP, OSPF, IS-IS and EIGRP. 

 

Some router are also capable of running IPv6 routing protocols such as RIPng, OSPFv3 and IS-IS (forIPv6). EIGRP (for IPv6) is not supported in OPNET Modeler 16. Figure 14 shows the available IPv6 protocols in OPNET suite.

Node models

LAN Workstation: A client side workstation that use service provided by the network servers

LAN: a number of workstation inter-connected together via internal switch.

Server: a station that provide a special functionality or service (e.g. FTP server)

Router: represents a Cisco 7200 router that will run the several dynamic routing protocols.

Switch: represent Cisco 2900 switch used to connect routers with LANs.

Links: represent a communication link that inter-connects network nodes.

Application Definition: defines the different application such as HTTP, FTP, VoIP, etc…

Profiles Definition: describe the activity patterns of the user in terms of the used applications over a specific period of time. Some LAN maybe runs a particular profile while other workstations are running another profile; each has its own defined application.

Link Failure/Recovery object: used to fail links during the simulation in a given times as well as recovering them afterward.