Performance comparison of shared Ethernet and switched Ethernet (Opnet Network simulation Experiment)

First, the purpose of the experiment

Compare the performance of shared Ethernet and switched Ethernet over different network sizes.

Second, the experimental method

Use Opnet to create and simulate a network topology, and run to analyze its performance.

Iii. contents of the experiment

3.1 Experiment settings (network topology, parameter settings, data to be observed, etc.)

Step 1: Download the It Guru, create a new project, and name it hub_switch_project.

Step 2: Create a scene--hub_scenario. Later we will add another scene--switch_scenario to the same project.

Step 3: Select the default model library and the default scene dimension.

Step 4: Drag and drop 5 Ethernet workstations and a server into the site, and connect them with a central hub. Topologies such as:

Step 5: Add the configuration file and the application node.

Step 6: Open the Application node and the Properties window select "Default" for the line defined by the application. This allows all standard applications (HTTP, FTP, e-mail, database, and so on) to be used in our network topology.

Step 7: We will create a customized profile--web application User profile-We call this summary "Web user". In the edit box for profile-Properties window. Such as:

Go to Profile Configuration–it is highlighted Above–and select "Edit". A new mini-window opens, and the Profile configuration table Window–select row #1. As shown in:

The current profile name is renamed "Web User". Change the recurrence column below the entry to "unlimited". The application chooses edit, when a new window pops up, this is a new application window, we will select an application that will be related to the Web user profile. Select Line 1th and select "Web browsing heavy." Such as:

Close all open windows. At this stage we have created a new profile called "Web User" that can be applied to our topology.

Step 8: Apply the configuration file "Web User" to all workstations. This makes it easy to select the following five workstations by holding down the control key, and then make the appropriate selection to change the Properties window for any one of the workstations. You must make sure that "Apply changes to Selected objects" is selected in the Properties window of the workstation. The profile is applied when the application tree is opened for the first time. The line entry that supports profile selection editing in the application. Open a new Window--supported Profiles window and select Web user on line 1th.

Step 9: Configure the server. Select Edit attributes, open the application tree, and go to apply row support services. Select Edit. A new window opens the Applications Services Table window. Select Line 1th and select Web browsing heavy. This stage is set on this machine essentially a generic HTTP server.

Step 10: Select the statistics that should be collected during the simulation run. In the menu bar Select Select Choose individual Statistics. In the Results window, select tree "Node Statistics".

Step 11: Run the Simulation and select "Configure/run des simulation" under the Des menu.

Step 12: Check the results via des menu, results->view statistics. such as:

Step 13: Find a useful comparison now let's create another scene under the same project. This makes it easy to replicate the work area of the entire topology center scene and subsequent changes to the copy in the new scene. Depending on the scene menu and select Copy Scene. The name is "switch scenario." A new workspace opens a repeating hub scenario. Select an Ethernet Switch Object window (click the Open Object panel button if it is not visible) and replace it with a hub.

Step 14: Run under the same data and switch the scene. By selecting the scene selection in the drop-down list, we can compare them in the same chart. Different data uses different colors. Check all the performance data.

3.2 Experimental Data

Node_0:

Node_4:

Contrast:

3.3 Experimental Data Analysis

By analyzing the object Response time and page Response time, you can see that both devices are fluctuating around a certain average value. However, compared to switch, the average response time of the hub is less and the delay is lower.

3.4 Question Discussion

1.In which scenario is there greater page loading delay? Explain why.

A: As you can see through page Response, when using a hub in a small LAN, page Response time is less than the switch. This is because the hub works in the physical layer, and the switch is working at the data link layer, which needs to be exchanged for processing, so the working time is relatively long.

2.Are there any worthwhile differences in any of the other statistics that were collected? Explain Why one would expect differences or identical results.

A: In other collected experimental data, User Cancelled connections is unimportant information.  But like page Response time is a valuable information, you can analyze the performance of each network through the comparison of two scenarios. The data produced in the experiment will clearly show the effect on the experiment under different preconditions, but the same result is easier to prove some consistency problem, or the change of some condition will not affect the result.

Four, the experiment content two

4.1 Experiment settings (network topology, parameter settings, data to be observed, etc.)

Step 1: Create a new project named Medium_hub_switch. Creating an empty case is called "Hub_scenario." ”

Step 2: Select a different model library from the default, and select Ethernet from the drop-down list in the model library. A list of new objects should appear.

Step 3: We will create a subnet.

Step 4: Double-click the subnet icon-this will open a new web window-the subnet window.

To create a new subnet. Such as:

Step 6: Click OK, it should automatically generate 14 workstations connected to the hub. Such as:

Step 7: Create another 14 workstation with hub hub connected to the topology subnet.

Step 8: Add the Application profile icon to the workspace and create a profile named Web user. Apply this profile to all workstations.

Step 9: Drag Bridge to the workspace and connect to the two hub hub. Such as:

Step 10: Return to the upper desktop.

Step 11: Copy the subnet icon and paste it, at least three to four--about 80-120 workstations. We can only create no more than 80 nodes here because of a version problem.

Step 12: Add a hub to the workspace and connect to the subnet's bridge via the Base T data cable.

Step 13: Set up the server.

In the Step 14:des menu, select Select Statistics and select Page Statistics time delay HTTP client.

Step 15: Run the scenario.

Step 16: Review the collected statistical data.

Step 17: Copy the current scene to a new scene called Switch_scenario.

Step 18: Replace the bridges under each subnet with a switch.

Step 19: Replace the top-level hub with switch.

Step 20: Select the same statistics.

Step 21: Review the results.

4.2 Experimental Data

Hub_scenario:

Switch_scenario:

Contrast:

4.3 Experimental Data Analysis

Analysis of the above experimental data can be obtained, in the medium-sized Ethernet, switching Ethernet performance is better than the shared Ethernet, but because the experimental simulation of the number of working host, the difference is not very large.

4.4 Question Discussion

(1) in which scenario is there greater page loading delay? Explain why.

A: In the Switch_scenario scenario, the page load time delay is longer, because the use of a switch device, so that data transmission needs to encapsulate the packet processing MAC address, so the corresponding cost increases. However, because the number of working hosts is small, the difference between the two is not very large.

(2) Is there any worthwhile differences in any of the other statisticsthat were (optionally) collected? Explain Why one would expect differences oridentical results.

A: No, the data produced in the experiment will clearly show the effect on the experiment under different preconditions, but the same result is easier to prove some consistency problem, or the change of a condition will not affect the result.

(3) re-run the Switch_scenario by replacing switches in the Subnetswith bridges in all the subnets. What is happens to the performance Then–to thepage delay statistic?

A: After replacing switch in all subnets, you can find that the page response time does not change much through data analysis and graphics discovery.

(4) re-run the Hub_scenario by replacing, the central hub in the top-levelnetwork with a bridge. What happens to the performance then–to the page delaystatistic?

A: After making changes, you can see that the page response time is significantly smaller than the previous.

(5) re-run the Switch_scenario by replacing the top-level switchwith a hub, but retaining the switches of all subnets. What happens to theperformance then–to the page delay statistic?

A: After making changes, you can find that the page response time is significantly larger than the previous

V. Conclusion

In small-scale Ethernet, the integrated performance of shared Ethernet is better than switching Ethernet. From the above analysis, it can be concluded that the transmission delay of switched Ethernet is larger because switch in switched Ethernet will encapsulate the packet, parse and parse the MAC address in the packet and forward to the target port, so these operations will create a certain delay. Instead, the hub does not do any processing of the packets but sends them directly, so the delay medicine is smaller. Therefore, in small-scale Ethernet, the integrated performance of the shared Ethernet is better than the switched Ethernet.

However, the shared Ethernet is a broadcast-based way to send data, because the hub does not recognize the frame, so when the data transfer, the direct data transmission and not packet processing, that is, all the hosts in the network can receive this data. So when the nodes in the network are increasing, the sharing ethernet that transmits the data by broadcasting is prone to the information blocking and conflict, which leads to the decrease of work efficiency.

To sum up, although in small and medium-sized Ethernet, the performance difference between the two, or shared Ethernet is better than the integrated switch Ethernet, but because of the increase in network size, the probability of conflict increased greatly, the use of switched Ethernet in large-scale Ethernet can improve network performance.

Performance comparison of shared Ethernet and switched Ethernet (Opnet Network simulation Experiment)