Introduction
OPNET Modeler can be used in order to create and analyze network models; therefore, it can do organizational scaling to deal with a specific ‘what if’ problem. In addition, use of Project Editor is necessary in order to create a small network, gather and collect statistics about network effectiveness and performance, run a network simulation, and analyze the received results and statistics. Moreover, designing a new network can be made easier if a group of several scenarios is combined in order to explore different network aspects.
In this lab report not only Loads and Delays on the first floor and expanded networks will be discussed, but also the difference between profile definition and application definition, and the traffic on two types of database accesses, which are Light Data Access and Heavy Data Access.
Objective
The objective of this lab practice is to create a solution for the case when a company plans to extend or increase its small intranet network. At the moment, the first floor of the company has a star topology network, hence, they are willing to expand and add an additional star topology network on another floor. Therefore, there is a need to test the ‘what-if’ scenario in order to ensure that when adding the second network, the additional network load will not cause the whole network to fail.
Once the network configuration, types of nodes within the network, and the types of links which connect the nodes to each other are selected, as a result, Rapid Configuration builds the network in one action. There are several additional steps that involve creating the first-floor network with Rapid Configuration:
In the dropdown menu select Star then click Next.
Set the Centre Node Model to 3C_SSII_1100_3300_4s_ae52_e48_ge3. This is a 3Com switch.
Set the Periphery Node Mode to Sm_Int_wkstn, change the node Number to 30.
Set the Link Model to 10BaseT.
Furthermore, to specify where the new network will be placed:
Set the X center and Y center to 25.
Set the Radius to 20.
Click OK.
Figure 1The First Floor Network
There are several steps required for creating the network objects:
1. Open the Object Palette.
2. Find the Sm_Int_server object and drag it into the workspace, and change its name to node_31.
3. Find the 10BaseT link object to connect the server to the star network.
4. Find the Sm_Application_Config object and drag it into the workspace, and change its name to node_32.
5. Find the Sm_Profile_Config object and drag it into the workspace, and change its name to node_33. Fig1.2 shows the completed first floor network.
Figure 1.2 Completed First Floor Network
After creating the first floor network, it not only has to be expanded, as one of the objectives of this practical lab states, but also the expanded network operation under heavier load has to be verified. To optimize the workflow, it makes sense to duplicate the scenario because the networks on two floors have to be similar. In order to do so, the next steps are to be performed:
1. Choose Scenarios > Duplicate Scenario.
2. Enter expansion as a name for the new scenario.
3. Click OK.
The second floor segment looks just like its first floor counterpart, but it needs to have its own server. So, we need to perform the following steps:
1. From Topology select Rapid Configuration.
2. Pull-down the menu and select Star then click next.
1. Set the Centre Node Model to 3C_SSII_1100_3300_4s_ae52_e48_ge3. This is a 3Com switch.
2. Set the Periphery Node Mode to Sm_Int_wkstn, change the node Number to 15.
3. Set the Link Model to 10BaseT.
4. Set the X center to 85 and Y center to -10.
5. Set the Radius to 20.
6. Click OK.
There is one more thing to do in order to join the two networks together, which are described below. Fig1.3 shows the two networks together when the steps below are completed:
1. Click on Tool button to open the Object Palette.
2. Expand Cisco 2514 folder and drag the Cisco 2514 router icon between the two networks in the workspace.
3. Expand the Link Models folder select 10BaseT link icon to link the center of each of two 3Com switch to the center of the Cisco router.
4. Close the Object Palette.
5. Select File > Save.
Figure 1.3 The Connection between the Networks
Result and Discussion
When the first floor and the expanded network statistics are ready to be collected, compared and discussed, Node and Process Editors have to be explored first because they play an essential and substantial role in creating a network at the first place. The Node Editor is used to create node models which describe the internal flow of the data within the network object. Node models are made up of one or more modules connected by packet stream or statistic wire, whereas the Process Editor creates process models that are represented by a state transition diagram (STD) that describes the behavioral logic of a module in a node model.
Ethernet loads and delays on the server of light data access and heavy data access are two substantial reasons for this lab practical to be distinguished and differed, when it comes to testing the second network scenario added to the first network, hence Fig1.4 shows the loads on the server of light data access for both networks scenarios. The first floor network is represented by the red line and at its peak load on the server is about 5,500 bits per second. The blue line represents the expanded network, and it is evident that the load rose up roughly twice as it was before adding the second network scenario, and is now at about 9,000 bits per second.
Figure 1.4 Loads On Server Of Light Data Access For Both Networks
In terms of the Ethernet delay for the light data access, the delay would exist, but it will be insignificant.
The examination of the delay for both network scenarios is shown in Fig1.5 and it proves that, in case of the light data access network, the delay occurs, but at very small rate. The red graph stands for the first floor network scenario, whilst the blue graphs represents the expanded network scenario, and it is clearly visible that both delays start within a rate of slightly higher than 0.25 milliseconds and reach a steady level of 0.4 milliseconds, showing no significant changes in Ethernet delay of the network.
Figure 1.5 Ethernet Delay Of Light Data Access For Both Networks
On the other hand, in term of heavy data access networks’ Ethernet loads and delays are expected to be greater than in case of the light data access networks for both the original and expanded network. Fig1.6 shows the Ethernet load of heavy data access for both networks.
Figure 1.6 Ethernet Load Of Heavy Data Access For Both Networks
The graph on Fig1.6 illustrates Ethernet load on first floor network that is represented by the red line and at the peak it slightly overshoots 5,000 bits per second. From the same graph it can be seen that when the second network is added, the load increases by almost five times with its peak hitting 25,000 bits per second.
Another aspect that needs to be mentioned in this report is the Ethernet delay, because in case of heavy data access networks it obviously is going to be greater after the addition of another network. As a result, Fig1.7 of this lab experiment demonstrates the evidence of a greater delay that occurs after the expansion of the first floor network.
Figure 1.7 Ethernet Delay Of Heavy Data Access For Both Networks
Last, but not least, the delay across the network is experienced by all users, hence, in order to understand where the delay occurs, it is recommended to separate voice network form data network, as the size of one data packet is 1500 bytes, whereas one voice packet is 80 bytes in size.
There are four types of delays that might affect the performance of the network for real-time applications:
1. Transmission Delay: known as Length of packet / Rate of the link = L / R
2. Propagation Delay: Defined as Distance / Speed = D / T, where speed of free space is 3 x 10^8 speed of copper or other material is 2 x 10^8
3. Queuing Delay: Defined as the time data takes to be served, this will be discussed more in the M/M/1 lab report
4. Processing Delay: For instance, typing on the keyboard, but it is a very small delay due to the high processing speed of computers nowadays.
One of the main aspects of this lab experiment is being able to distinguish and tell the difference between a profile definition and an application definition.
Profile Definition defines what the user’s behavior is.
Application Definition: all applications will produce different traffic profiles e.g.
• Light Database Access is 16 bytes every 30 s: this gives 16x8 / 30 = 4.27 bps
• Heavy Database Access is 32764 bytes every 12 s
In order to calculate the transfer rates of the heavy data access the formula below has to be used:
Transfer rate = (Transaction size in bits x8 / transaction time) x (number of interfaces). Transfer rate = ((32764 ×8) /12) × 45, where 45 are numbers of interfaces in both networks: 30 on the first floor and 15 on the expanded network.
Traffic on the network = 983040 bps.
Conclusion
OPNET has illustrated the effectiveness and the performance of the network in a ‘what-if’ scenario, when the first floor network was expanded. Due to the expansion of the network, load and delay were increased on both application profiles, especially on the heavy data access. Moreover, there are four types of the delays:
• Transmission Delay
• Propagation Delay
• Queuing Delay
• Processing Delay
It is recommended that delay on voice IP network is 150ms, whereas delay on gaming network is between 50 - 90ms.
Another aspect that OPNET has illustrated is the difference between a profile definition and an application definition.
Profile Definition: known as what the user are doing.
Application Definition: known as what types of databases users are using, hence there are two types of databases:
• Light Database Access is 16 bytes every 30 s: this gives 16x8 / 30 = 4.27 bps
• Heavy Database Access is 32764 bytes every 12 s