' The College Network - Group Project(with Comnet)'

  • CSCI 422
  • Tutor : Dr. Adda
  • by : Anthony Lopez-Vito, Kwang-Joon Kim, Naznin Todiwala, Dimiter Tsotsorkov.

    Objective

    The purpose of this project is to know how the computer network is designed and installed at our college. By simulating college network with Comnet, we learn how to design and simulate with Comnet III, and suggest improvement of present network with result of the simulation.

    Specification

    Present Network

    Richmond College network is segmented into 4 Local Area Network; Richmond, Atlantic, Pacific, Young Street. And these separated LANs are connected by Megastream which is 1Mb/s. There are Library system, Novix Server, Tetra Account, Student Information System and CD ROM Server in some LANs. These, however, are hardly used comparing to MS office, e-mail, and Internet. Therefore, those are ignored but, only attached each LAN Server, computer group, and e-mail gateway.

    FACTS

    There are 312 workstations which are divided into different LANs.

  • Richmond = 96 stations.
  • Atlantic = 96 stations.
  • Pacific = 48 stations.
  • Young Street = 72 stations.

    These numbers of workstation are set on Computer Group Parameters with 500 Mb storage for each station. Each computer group request e-mail, e-mail check, Web and file to both local and global network.

  • 4 Servers - Richmond, Atlantic, Pacific, Young Street.
  • e-mail Server - Pacific .
  • Home Drive for students(h:\) - Pacific.
  • Bandwidth of Megastream - 1Mb voice / 1Mb data.
  • Bandwidth to ULCC - 64kbps.
  • Bandwidth of LANs - 10Mbps

    Implementaion

    It has been suggested to run two different network, busy hours and not-busy hours for both present and improved network in order to see the channel utilisation of network with different amount of packet.

    Not Busy Network

    E-mail is set to a uniform distribution with a minimum of 300 seconds and a maximum of 900 seconds for the interarrival. Also, every e-mail messages are designated to mail gateway only.

    E-mail check is set to 60 bytes for probability distribution and, again, designated to mail gateway only. Set to 900 seconds for interarrival. Web Request is set to uniform distribution with a minimum of 60 seconds and a maximum of 300 seconds for interarrial, and set to weighed list to ULCC. File Request contains as Figure 1. It is set to a uniform distribution with a minimum of 300 seconds and a maximum of 900 seconds for the interarrival. First it request Office program from LAN server since it is the most common used program in our college, and request file from home drive(h:\) which is located at Atlantic server for all students then terminate its job after save the file used. These commands are set by local computer group for office request, office wait, office processing and application processing, and file request, file request wait and save request by Global Command Repertoire.

    Busy Network

    E-mail is set to a uniform distribution with a minimum of 240 seconds and a maximum of 600 seconds for the interarrival. Also, every e-mail messages are designated to mail gateway only. E-mail check is set to 60 bytes for probability distribution and, again, designated to mail gateway only. Set to 900 seconds for interarrival. Web Request is set to uniform distribution with a minimum of 45 seconds and a maximum of 200 seconds for interarrial, and set to weighed list to ULCC. File Request contains as Figure 1. It is set to a uniform distribution with a minimum of 150 seconds and a maximum of 500 seconds for the interarrival. First it request Office program from LAN server since it is the most common used program in our college, and request file from home drive(h:\) which is located at Atlantic server for all students then terminate its job after save the file used. These commands are set by local computer group for office request, office wait, office processing and application processing, and file request, file request wait and save request by Global Command Repertoire.

    Figure 1

    Figure 1 shows what has been added on the file request command. After all, the LAN looks like figure 2.

    Figure 2

    Whenever, a workstation ask for MS Office program, it request to its own LAN server so, each LAN server response to its own computer groups only as figure 3.

    Figure 3

    Application source of Office Response is set to received message option for schedule by and to the received message option with requiring one message with the test ‘Office Request’ to trigger the application source.

    While all servers need to answer ‘MS Office Request’ only, the Pacific server has to answer for ‘file request’ and ‘save request’ to all workstations within college network. Therefore, the Pacific server has ‘file request response’ and ‘save file’ for all LANs rather than only ‘MS Office response’ as figure 4.

    Figure 4

    Command sequence of ‘File Request’ is set as Figure 5. Any packet ‘file request’ simply trigger the ‘file request response’ and so it can answer the request from workstations.

    Figure 5

    ‘Save file’ is very similar to File Request but much simpler as figure 6.

    Figure 6

    There are one Mail Gateway inside the Pacific LAN. It is triggered by ‘e-mail check’ of any workstations and set to uniform distribution with a minimum of 3 seconds and maximum of 5 seconds for the delay. Also, set to a normal distribution with a mean of 40000 bytes and a standard deviation of 10000 bytes.

    Mail Gateway

    Web is responded through ULCC. Web page read is set as the tutorial table(p73). ‘Web Response’ command is set as figure 7.

    Figure 7

    Web Response

    Improved Network

    It is very easy to improve any network if finance is not the matter. However, money is the most important issue of any organisations when they want to install or improve network. Therefore, improved network has been done with existing hardware only.

  • Richmond College LANs are switchable from 10mb to 100mb.
  • Even though Megastream is 2MB for both data and phone, it has been divided into 50% between buildings. It would be wasting to giving 1mb just for phone although it could work with 0.5mb. Therefore, I divided to 75% to data and 25% to telephone.
  • The line to ULCC has been improved from 64kb to 2mb.

    Implementation;

    As the present network, improved network has been run for two different amount of message.

  • Not busy network same as present network……

  • Busy network same as present network……

    Simulation Result

    Conclusion
    Four reports shows that what are the channel utilisation for each network but, less channel utilisation than expected. It is, perhaps, we did not set the same ‘uniform distribution’ as it is on real network. However, it is fact that percentage of channel utilisation of improved network decreased from present network. It implies that there is less delay for message but channel is used less time than present. In other words, if the channel utilisation goes up then message delay goes up as well. It is depends of the users which kind of channel they prepare. However, it is true to say that there are so much traffic on our college network and doesn’t cost much to switch to higher bandwidth, it would be better to use improved network.