OPNET_LabManual

Lab Manual for OPNET Training Using this book In each section, there is some information about what you are doing and why, and some step-by-step instructions in bold type. There are also screenshots of the lab at various stages. For each section, read through the lab once to see what you will be doing, then go back and follow the instructions. There are screenshots above some sections that show what your project should look like. When you are done with the section, if the rest of the class is still working, we suggest that you look at the Online Documentation (under the Help menu) and search for relevant topics or topics of interest to you. IT Guru Day 1 Lab 1 : Creating a Topology Lab 2 : Adding Traffic Lab 3 : Choose Statistics and Run Simulation Lab 4 : View Results Lab 5 : Create Additional Scenarios IT Guru Day 2 Lab 6 : Building Network Topologies Lab 7 : Application Response Time Engineering Lab 8 : Traffic Engineering using MPLS Lab 9 : Simulation Methodology for the Analysis of QoS OPNET Modeler Day 1 Lab 10 : Creating Node Model Lab 11 : Creating Link Model Lab 12 : Completing the Bank Network OPNET Modeler Day 2 Lab 13 : Expanding the Bank Network Lab 14 : Creating Process Model Lab #1: Creating a Topology In order to represent your network’s topology, you must first gather information about the aspects of your network you wish to model. Below is some information about the company and its network. Lansing Hotel Reservation Services is a hotel reservation company headquartered in Dallas, Texas. Lansing employees are experiencing high delays using the company’s proprietary hotel reservation application, and customers are threatening to use other services if the problem is not fixed. Your task is to identify the problem and propose a solution. Lansing’s current network consists of four offices, each containing 28 interconnected workstations. The offices are connected via 56K lines. The Application Servers, which all users access to run Lansing’s proprietary software (as well as other important applications), are located in the company’s Dallas HQ. Create a new project: · Select File / New / Project; press “OK” · Set Project Name = Hotel_reserve · Set Scenario Name = original_network; press “OK” Answer questions in the Startup Wizard: · Initial Topology box: click on “Create Empty Scenario”; press “Next” · Choose Network Scale box: click on “Choose From Maps”; press “Next” · Choose Map box: click on “USA”; press “Next” · Select Technologies box: scroll down and include “Hotel_reserve_palette” by changing the “No” to “Yes”; press “Next” · Review box: press “OK” Change view to the one shown on the previous page: · Click on the Zoom button on the toolbar · Click and drag a rectangle over the Midwestern portion of the US Open the Object Palette: · If it’s not already open, either click on the Object Palette button, or select Edit / Open Object Palette Deploy subnet to Minneapolis, and change its grid properties so that the scale is smaller: · Click and drag a subnet icon from the palette onto the workspace, placing it on Minneapolis · Right-click to cancel placing additional subnets · Right-click the subnet and select Advanced Edit Attributes · Change the “x span” and the “y span” to .002 Change to subnet view: · Double-click on the subnet Change Grid Properties: · Select View/Set Grid Properties · In Map Grid Properties dialog box, change units from “Degrees” to “Feet”, change resolution to “8”, and change division to “10” · Close the Map Grid Properties dialog box Create a copy of this empty subnet in Dallas: · Exit out of the subnet by pressing the “go to next higher level in network hierarchy” button, or by right-clicking in the workspace and choosing “go to parent subnet” · Copy the subnet by selecting it and hitting “Control-C” · Paste another subnet on Dallas by hitting “Control-V” on the keyboard and left clicking above Dallas In Minneapolis, rapidly configure a LAN consisting of 28 workstations connected to a 3com_SuperStack_II_Switch_3900 in a star topology: · Enter the Minneapolis subnet · Select Topology / Rapid Configuration; change “Bus” to “Star”; press “OK” · Set Center Node Model = “3com_Superstack_II_Switch_3900” · Set Periphery Node Model = “ethernet_wkstn” · Set Link Model = “100BaseT” · Set Number = “28”. Press “OK” Add a router to the LAN: · Click on the “CS7505_Router” router in the Object Palette, and drag it into the workspace · Right-click in the workspace to stop adding routers Change name of the router to “cisco_router”: · Right-click on the router; choose set name; type “cisco_router”; click OK Connect the Router to the LAN: · Click on the “100BaseT” link in the object palette · Left-click on the router and then left-click on the hub · Right click in the workspace to deselect the link · Your subnet should now look like the following picture Go back to WAN view of network: · Right-click in workspace and select “Go to Parent Subnet” Copy the Minneapolis subnet to Detroit, and Atlanta: · Select the Minneapolis subnet; copy it using Edit / Copy or Ctrl-C · Choose Edit / Paste or Ctrl-V; left-click on Detroit · Repeat on Atlanta Name the subnets “Minneapolis”, “Detroit”, “Atlanta” and “Dallas”: · Right-click on the subnet; choose Set Name · Enter the appropriate name Create the topology in Dallas: · Enter the Dallas subnet · Deploy 6 Sun_Enterprise_10000_server models from the Object Palette · Deploy a 3com_SuperStack_II_Switch in the middle of the servers, and connect the servers to the switch with 100BaseT links · Place a CS7505_Router next to the switch and connect the two with a 100BaseT link · Set the name of the new router to “cisco_router” · Right-click on the workspace and choose “Go to Parent Subnet” Save the project: · Click File / Save Place an IP Cloud in the middle of the 4 subnets: · Click on the Object Palette icon to open it (again, this is only if the Object Palette isn’t open already) · Drag the icon labeled ip32_cloud onto the workspace · Connect the Dallas subnet to the cloud with a PPP_DS1 link · When prompted for which node within the subnet to connect the link to, choose the node name ending in “cisco_router” · Right-click in the workspace to stop placing DS1 links · Connect the cloud to the other 3 subnets using a PPP_56K link · Again when prompted for which node within the subnet to connect the link to, choose the node name ending in “cisco_router” · Click File / Save · Verify that your network topology looks like this: End of Lab #1 Lab #2 – Adding Traffic A simulation with no traffic would be quite boring. Let’s add some traffic to the network topology. The traffic you are most interested in is the Database traffic generated by the Hotel Reservation Software. You will model this traffic explicitly using IT Guru’s pre-defined application traffic. There is, however, other traffic flowing to and from the server farm in Dallas. You will simplify this other traffic by modeling as conversation pair traffic. Lansing has run some traffic captures and stored them in a text file. You will need to import this text file. Import background routed traffic into the simulation: · From the traffic menu, choose Import Conversation Pairs / From Spreadsheet · Choose the traffic file called: Hotel_reserve_traffic.txt · Choose “replace all existing traffic” and hit OK Optional: To view the background traffic, use the Conversation Pair Browser under the traffic menu. Select a source and a destination, then right-click on the destination and choose “view traffic”. Deploy an Application Config object: · Drag and drop it from the Object Palette into the workspace Edit the Application Config object so it sends traffic representative of Lansing’s reservation software: · Right-click the Application config node, choose Edit Attributes · Click on the Value column of the Applications Definition attribute, and choose Edit…. · Click on the box that says “0 rows” and change the 0 to a 1 · For the Name value, type “DB User”, and for the Description, choose edit · In the next box that appears, click on the value column next to Database, and choose edit · In the next dialogue box, fill in the following information: Note: To choose a distribution for transaction interarrival time and transaction size, you will need to change the “Special Value” from None to Not Used. · Once your attribute list matches the one above, hit OK until all the boxes are closed Next, you will need to create a user profile for someone who uses this database application: · Deploy a Profile Config node from the Object Palette to the workspace · Right-click it and Edit Attributes · Change the Profile Configuration attribute from None to Edit… · Click on the box that says “0 rows” and change the 0 to a 1 · Change the Profile Name to DB Client · Change the Start time to a uniform distribution between 200 and 400 seconds. (This gives the routing protocols enough time to find a path through the network, it also gives you a wide enough range so that all the workstations don’t send their first query at once.) · Click on the Applications column and choose edit · Click on the Name column, choose “DB User” · Click OK until all dialogue boxes are closed Now you have configured the application and the user profile. The next step is to assign the user profile to all the workstations in the network. Select all workstations to set their user profiles at once: · Double-click on the Atlanta subnet and right-click a workstation · Choose “Select Similar Nodes” · Right-click on the workstation again and edit attributes · First, check off the box at the bottom of the window for “Apply Changes to Selected Objects” · Next, change the value of the Application: Supported Profiles attribute to “Edit” · Add one row and click on the name column · Choose “DB Client” · Click OK until the dialogue boxes are closed The servers are pre-configured to support this application, so you will not need to modify them. · Save your project End of Lab #2 Lab #3: Choose Statistics and Run Simulation First, you must select some statistics to collect: · From the Simulation menu, select Choose Individual Statistics · Click on the plus signs next to the categories to expand them · Choose the following statistics by clicking on the gray box next to the statistic: · Global Statistics / DB Entry / Response Time (sec) · Global Statistics / DB Query / Response Time (sec) · Node Statistics / Server DB Query / Task Processing Time · Link Statistics / point-to-point / utilization <- · Link Statistics / point-to-point / utilization -> · Click OK Next, you must configure and run the simulation to run for one hour: · Select Simulation / Configure Simulation · Set the duration to one hour · Press “Run” End of Lab #3 Lab #4: View Results Create panels for DB Query and Entry response times and their averages: · Select Results / View Results · Check the box next to Global Statistics / DB Entry / Response Time (sec) · Press “Show” · In the View Results window, change the filter value from “As Is” to “average” · Press “Add” and click on the graph that was just created · Uncheck the box next to DB Entry / Response Time (sec) · Repeat the steps above for DB Query / Response Time (sec) · Verify that your graphs look similar to the graphs below – you may have peaks and valleys in different places, but the range of values should be similar · Save the project This Database application is taking around two seconds to deliver a response. Try to discover what is causing the slow response times. First, look at the server utilization to see if that is the culprit. · Select Results / Find Top Results · Choose Node Statistics / Server DB Query / Task Processing Time, click “Find Top Results” · Change “Statistics Stacked” to “Statistics Overlaid” · Click the Graph button and view the graph · Notice the processing time is quite low If the servers are not causing the bottleneck, perhaps the links are over-utilized. Check the most utilized links for bottlenecks: · Select Results / Find Top Results (the dialogue box should still be open) · Choose Link Statistics / point-to-point / utilization, click “Find Top Results” · Change “Statistics Stacked” to “Statistics Overlaid” · Click the Graph button and view the graph · Notice which of the links have 100% utilization for part of the simulation · Save the project In this section you diagnosed the slow response times – more bandwidth is needed on the long-distance links. End of Lab #4 Lab #5: Create Additional Scenarios Now it’s time to fix the network bottleneck. Since the links are over-utilized, try upgrading all the 56K links to DS1 links. First, you’ll want to create a duplicate scenario in which to make this change. · Select Scenarios / Duplicate Scenarios · Name the new Scenario “DS1” · Right-click on the link connecting the Detroit Subnet to the IP Cloud and “Select Similar Links” · Right-click on this link again and select “Edit Attributes” · Change the Model attribute from “PPP_56K” to “PPP_DS1” · Check the box that says “Apply Changes to Selected Objects” and press “OK” Run the Simulation: · Press the simulation icon and press Run, or select Simulation / Run Simulation Create a panel containing both DB Query and DB Entry Response Times: · Choose Results / View Results · Check both Global Statistics / DB Query / Response Time (sec) and Global Statistics / DB Entry / Response Time (sec) · Change “Statistics Stacked” to “Statistics Overlaid” and click Show At a glance you can see that there is a significant reduction in response time. Let’s see if you can get a further improvement by upgrading to DS3 links: · Select Scenarios / Duplicate Scenarios · Name the new Scenario “DS3” · Right-click on the link connecting the Detroit Subnet and the IP Cloud and “Select Similar Links” · Right-click on it again and select “Edit Attributes” · Change the Model from “PPP_DS1” to “PPP_DS3” · Check the box that says “Apply Changes to Selected Objects” · Press “OK” Run the Simulation: · Select Simulation / Run Simulation Compare the DB Entry and DB Query Response Times for each scenario: · Select Results / Compare Results · Check the box next to Global Statistics / DB Entry / Response Time (sec) · Change the filter to “average” · Click “show” · Uncheck the box next to Global Statistics / DB Entry / Response Time (sec) · Repeat these steps for Global Statistics / DB Query / Response Time (sec) Generate a Web report for the “DS3” scenario, and then launch it: · Select Results / Statistic Report / Generate Web Report and press “OK” · Select Results / Statistic Report / Launch Last Report View DB Query Response Time: · If you have more than one Web report, you will start at your Simulation Reports Home Page and would then have to select the simulation report you wish to view · If this is the first report ever generated, you will bypass this home page · Find and view the results for Report: User Selected / Global Statistics / DB Query / Response Time · Exit the browser · Save the project End of Lab #5 Lab #6: Building Network Topologies Background Reduce simulation time by selecting effective topology scale and traffic types while maintaining accuracy of database response time statistics · For the purposes of this lab, it is assumed that on average an IP packet is about 1000 bits in size Instructions Open “1203_Lab1” project 1. Use Hide/Show All Graphs button to view saved results 2. Observe that DB traffic represents less than 10% of total traffic on network 3. Note that all sites contain explicitly defined workstations 4. Switch to second scenario, “Explicit_Web_Traffic_Only” 5. Use Hide/Show All Graphs button to open saved results 6. Each graph shows the average throughput on a WAN link in either direction 7. Note the amount of traffic that flows between each city · Example: Traffic between Dallas and Denver · Observe steady state traffic in both directions · Dallas -> Denver = 2200bps · Denver -> Dallas = 100,000bps 8. Complete the matrix below using the same technique as in previous step. Source Destination Avg throughput (bps) Atlanta.Router Denver.Router 6000 Atlanta.Router Miami.Router Atlanta.Router Tampa.Router 100000 Chicago.Router Denver.Router 12000 Chicago.Router Detroit.Gateway Dallas.Router Denver.Router 2200 Miami.Router Atlanta.Router 2000 Tampa.Router Atlanta.Router 2100 Washington DC.Router Pittsburgh.Router 2100 9. Switch to third scenario, “Aggregation_Explicit_and_Conv_Pair_Traffic” 10. In each site, aggregate workstations into 100Base-T LAN objects 11. Set the “Number of Workstations” attribute for each LAN object equal to the number of clients that were aggregated 12. Set all LAN objects to run Database profile only · Right-click on any LAN object and select “Select Similar Nodes” option · Edit the “Application: Client Supported Profiles” table · Database profile should be the only one in the list · Return to main attribute list · Enable “Apply Changes to Selected Objects” checkbox and press OK 13. Use results observed from step 9 to convert HTTP traffic into conversation pairs i. Use “Export Conversation Pairs” option in traffic menu to export existing traffic to spreadsheet with settings below: ii. Use throughput values from matrix in step 9 to fill in values in spreadsheet iii. Set “Avg Pkt Size” value to 1000 for all entries iv. Save file and exit from Excel v. Import modified file into topology using Import Conversation Pairs / Spreadsheet option in the Traffic menu making sure that existing traffic is overwritten 14. Run a simulation for two hour duration and note time to complete simulation (wall-clock time) 15. Compare results for “DB Query Response Time” statistic for all scenarios · Select “Compare Results” option from the Results menu · Select Global Statistics / DB Query / Response Time (sec) · Press “Show” button to plot results · Right click on panel and select Draw Style / Discrete option · Change “As Is” field in compare results window to “average” · Press “Show” button to plot additional results graph Optional Model the effect of twice as many users on the network. 1. Scale conversation pair traffic by 100% by using the “Scale Network Traffic” option in the conversation pair browser 2. At each site, double the number of workstations in each LAN object 3. Run simulation for two hours and note simulation run time (wall-clock time) 4. Compare database application response time for all three scenarios Conclusion · Aggregating workstations into LANs and modeling HTTP traffic as conversation pair traffic significantly reduced time to run simulation while maintaining accuracy of results. · Using methodologies to build network topologies and select traffic can allow you to run studies efficiently without sacrificing accuracy. LAB #7: Application Response Time Engineering Background Apply SMARTE to deploy application for EuroBank and meet the defined Service Level Agreement (SLA). · EuroBank has 12 branches located across Europe with a central office in London · Data center is located in Zurich along with a data warehouse server that replicates bulk information from central offices located across the world · New database application to replicate branch data is being deployed at all European branches characterized by an ACE model. · SLA for new application is defined as <25s response time 95% of the time · Effect of additional applications on network elements has been configured as conversation pair traffic and device loads Instructions 1. Use Hide/Show All Graphs button to show application response time and SLA compliance graphs