Overview
In this lab, we will compare the speed and accuracy of different traffic representations: explicit traffic, background traffic, and hybrid traffic. The network used in the lab is a model of a company that provides video-on-demand services to 100 users.
The company would like to introduce three classes of service for its clients: gold (ToS = 3), silver (ToS = 2) and bronze (ToS = 1). To provide differentiated treatment for the different service classes, Weighted Fair Queuing (WFQ) has been configured on the access router. In this lab, we will predict the delay for each class of service and compare the results obtained using the different traffic modeling approaches.

Objectives and Methodology

* Create a simple network with explicit traffic and run a simulation. * Replace explicit traffic with background traffic and rerun the simulation. * Replace background traffic with hybrid traffic and rerun the simulation. * Assess and compare the speed and accuracy of the three traffic-modeling approaches.

Explanation and Background
In the real world, one of the most important jobs a network manager can do is manage the traffic on the network. If the traffic doesn’t flow, the network is not exactly a credit to its operators. In addition, we now deal with many different types of traffic that are particular to certain applications and architectures, the most obvious examples being voice and video traffic.
In these applications, the number one demand is for low delay. Other applications may have different demands, such as high throughput, high reliability, or high security, which are more important then delay.
OPNET will very accurately model three types of traffic.

a. Explicit Traffic: This is the traffic we especially want to test and monitor. This traffic, be it TCP, HTTP, or UDP voice, is explicitly modeled, packet by packet, and we can retrieve all sorts of information on it. b. Background Traffic: Networks are not all explicit traffic, but also DNS, DHCP, printer blocks, and broadcasts. Although there are times we want to know specific things about some or all of this traffic (broadcasts, for example), we generally only want this traffic present in order to create a realistic load in the network simulation. c. Hybrid Traffic: This is the more normal state of having a mixture of both explicit and background traffic. Generally, however, most of the traffic will be background.
The important thing to remember is that not all traffic needs to be explicitly modeled, only the specific traffic we are interested in. In most cases, we want to see this traffic running under load, so we add background traffic to the explicit traffic to achieve the load we need/want.
Another side to this is how much time you want to spend doing the analysis. Because explicit traffic is modeled packet by packet, it takes processor cycles and time to simulate it.
This lab is intended to show you how to model all three types of traffic and test a set of circumstances against these types.

Procedure

Open Project

1) Select File/Open

“op_models\netw410\TrafficModeling.project\TrafficModeling.prj”

The Scenario “Explicit Traffic” should open.

Comparison of Traffic Modeling Approaches
Network Model 1. Source nodes (video servers) with different ToS (1, 2, 3) Interface with WFQ 2. 100 video users (MPEG4)—3 available service classes 3. Destination nodes

Part 1: Simulation With Explicit Traffic
Create explicit traffic using IP Traffic Flows
We will create three traffic flows representing the traffic downloaded by the three classes of clients.
In this scenario, all traffic will be modeled as explicit traffic.
1. Open the object palette.

2. Scroll down and double-click on the “ip_traffic_flow” demand from the demands palette.

3. Connect the IP demand between src_1 and dest_1 (in the same way you would create a link object).

4. Right-click on the workspace and select Abort Demand Definition to exit the demand operation. Close the object palette.
5. Right-click the demand object and select Edit Attributes.

6. Edit the traffic specification for both bits/second and packets/second values as described in the following steps.
a. Configure the Traffic (bits/second) attribute with a value of 6.0 Mbps from 0 to 600 seconds.
• The profile name may be different in your case.
• The graph is updated when you press Enter after entering the values.
b. Click OK to commit changes.

c. Configure the Traffic (packets/second) attribute with a value of 500 packets/second from 0 to 600 seconds.
• The profile name may be different in your case.
• The graph is updated only if you press Enter after entering the values.

d. Click on OK to commit the changes.
7. Set the Traffic Mix attribute to All Explicit.
8. Click OK to close the Demand Attribute dialogue box.
9. Copy the demand object.
a. Click on the demand object to select it.
b. Press Ctrl-C to copy the demand.
c. Press Ctrl-V to paste a copy of the demand in the same direction from src_2 to dest_2.
d. Press Ctrl-V again and paste a copy from src_3 to dest_3.

Configure the Type of Service on each demand
1. From the menu bar, choose Protocols > IP > Demands > Characterize Traffic Demands.
2. Set the Type of Service for the demands originating from WFQ_net.src_1, WFQ_net.src_2, and WFQ_net.src3 to Background (1), Standard (2), and Excellent (3) respectively.

3. Click OK.

Configure the packet size and packet interarrival time distributions
Next, we will configure additional traffic generation parameters, such as statistical distributions used for packet sizes and packet interarrival times. Note that these can be configured either individually for each demand (under the Traffic Characteristic attribute) or globally for all the demands (using the Background Traffic Config utility). We will use the latter approach in this lab and configure the distributions by modifying the global settings.
1. Right-click on the bkg_config node and select Edit Attributes.

2. Set the Packet Interarrival Time Variability to exponential.
3. Set the Packet Size Variability to constant.

4. Note that the average values for both distributions are already determined by the settings in the traffic profile and, therefore, appear as Auto_Calculated. The traffic volume of 500 packets/second in the profile translates into an average packet interarrival time of 0.002 seconds. The average packet size is also already determined.

5. Click OK to commit changes.

Choose statistics
We will monitor the queuing delay statistics for each service class on the outgoing interface of router B. This interface is the access interface to the core network and has WFQ scheduling enabled to provide QoS treatment to different service classes. We will also collect the packet end-to-end delay statistics for all the traffic flows.
1. Right-click anywhere in the Project Editor workspace and select Choose Individual DES Statistics.
2. Expand Demand Statistics by clicking on the (+) next to it and select the Packet ETE Delay (sec) statistics.

3. Expand Link Statistics and then expand point-to-point. Select both the throughput (bits/sec) and throughput (packets/sec) statistics.
4. Expand Node Statistics and then expand IP Interface. Select the Queuing Delay (sec) statistics.
5. Click OK.

Run the simulation
1. Click the Configure/Run Discrete Event Simulation toolbar button.
The simulation is set to run for 10 minutes of simulated time.
2. Start the simulation by clicking on the Run button. The simulation takes about one minute.
3. Close the Simulation Execution dialog box after the simulation finishes.

Results analysis
4. Right-click on the router_B ↔ IP_CLOUD link and select View Results.
5. Expand the point-to-point group and select the throughput (packets/sec) and throughput (bits/sec) statistics. Click Show. Lab Report Item 1. Capture a screen shot of the results screen and paste it into your answer sheet. Lab Report Item 2. What do the stats show about the traffic entering the cloud?
6. Close the Results Browser.
7. Right-click on router_B and select View Results.
8. Expand the IP Interface group and select the statistics below in the following order:
a. WFQ Queuing Delay (sec) IF10 Q3
b. WFQ Queuing Delay (sec) IF10 Q2
c. WFQ Queuing Delay (sec) IF10 Q1
9. Change the display mode from Statistics Stacked to Overlaid Statistics.
10. Change the As Is filter to average.
11. Click Show. Lab Report Item 3. Paste a shot of this screen into your worksheet.

Lab Report Item 4. What do you observe about the packet delay related to the ToS we assigned?
12. Close the Results Browser.

Part 2: Simulation With Purely Background Traffic
In the next scenario, we will replace the explicit traffic with purely background traffic. We will run the simulation and compare the results and the simulation speed.
Replace Explicit traffic with Background traffic
1. Choose Scenarios > Switch To Scenario > Background_traffic.
2. Right-click on the demand going from src_1 to dest_1 and choose Edit Similar Demands. You will see the Demand Properties screen.
3. Scroll right until you see the Traffic Mix field. 4. Change the Traffic Mix attribute to All Background as shown on all three demands.
5. Click OK. Save your scenario.

Run the simulation
1. Click the Configure/Run Discrete Event Simulation toolbar button.
2. Note that the simulation is set to run for 10 simulated minutes.
3. Run the simulation by clicking on the Run button. The simulation takes about four seconds. Lab Report Item 5. Why does this simulation run so much quicker than the first one?
4. Close the Simulation Execution dialog box after the simulation finishes.

Compare results with those of the previous scenario (with explicit traffic)
1. Click the Hide/Show Graph Panels button.
2. Load the panels with latest results by choosing DES > Panel Operations > Reload Data into All Panels.
3. The first graph panel displays the queuing delays for various queues at the access router_B obtained in this scenario.
4. The second panel compares the individual queuing delays between the two scenarios. Results analysis Lab Report Item 6. What do you observe about the queuing delays in both scenarios?
5. Click the Hide/Show Graph Panels button.
In the next scenario, we will see how we can overcome this shortcoming by using hybrid traffic (a mixture of background and explicit traffic).

Part 3: Simulation with Hybrid Traffic
In this scenario, we will configure the demands to use a mixture of explicit traffic (1%) and background traffic (99%). This way we will be able to obtain end-to-end statistics and still achieve significant simulation speed-up (compared to the purely explicit traffic scenario).
Note that the explicit traffic is only a small fraction of the total traffic. This is the recommended configuration. For flows with large traffic volumes, very small fractions, such as 0.01% or 0.1% explicit, should be used. Configurations such as 30% explicit + 70% background traffic do not result in significant simulation speed up and should be avoided.

Configure demands with hybrid traffic (1% explicit + 99% background)
1. Choose Scenarios > Switch To Scenario > Hybrid_traffic.
2. Right-click on the demand going from src_1 to dest_1 and choose Select Similar Demands. Note that the other two demands got selected.
3. Right-click on the demand again and choose Edit Attributes.
4. Change the Traffic Mix attribute to 1.0 % Explicit.
5. Click OK.

Run the simulation and compare results with other scenarios
1. Click the Configure/Run Discrete Event Simulation toolbar button.
2. Note that the simulation is set to run for 10 simulated minutes.
3. Run the simulation by clicking on the Run button. The simulation takes about eight seconds.
4. Close the Simulation Execution dialog box after the simulation finishes.
5. Load the panels with latest results by choosing DES > Panel Operations > Reload Data into All Panels.
6. Click the Hide/Show Graph Panels button.

Results analysis Lab Report Item 7. The first panel compares the queuing delays between the three scenarios. Paste a shot of this panel in your answer sheet.

Lab Report Item 8. How do you compare these results?

Lab Report Item 9. The second graph panel compares the packet end-to-end delays for the three traffic demands obtained using explicit traffic and hybrid traffic (no ETE delays were available in the purely background traffic scenario). Paste a shot of this panel into your answer sheet. Lab Report Item 10. How would you compare the results between the first and third scenarios?
7. Close the Results Browser window.
8. Save and close the project.
9. Complete your Lab Report and place it into the Dropbox.