SHEFFIELD HALLAM UNIVERSITY.

SCHOOL OF COMPUTING AND MANAGEMENT SCIENCES.

AN INTRODUCTION TO NETWORK
MODELLING AND NETWORK
SIMULATION USING COMNET III.

COMPUTER NETWORK EXPERIMENT NUMBER 4 (EXPT4).

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 1
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
TITLE

AN INTRODUCTION TO NETWORK MODELLING AND

NETWORK SIMULATION USING COMNET III.
AIM

To gain an understanding of network Modelling and network Simulation, using Comnet III software.

OBJECTIVES

1. You will be introduced to some basic concepts and terminology used for modelling networks.
2. You will be able to construct simple network configurations.
3. You will be able to interpret and modify parameters for sample network modelling objects.
4. You will be able to carry out simulations of the constructed networks.
5. You will be able to select and interpret the results of a network simulation.
6. You will gain a basic understanding of statistical probability and it’s use in network simulation.

PRE-REQUISITES

The Comnet III simulation software will run under the following PC operating systems.

 Microsoft Windows 3.1

 Microsoft Windows 95
 Microsoft Windows NT

You should be familiar with using the operating system installed on the machine you are using to run
the simulation software. (Note: Comnet III is installed under Windows NT 4.0 in the Networks lab).

EQUIPMENT REQUIRED

A networked PC Computer running an operating system specified below with at least the
corresponding machine specification. In addition, the machine must be licensed and registered to run
the Comnet III software.

 Microsoft Windows 3.1, 32-bit emulation software, 16Mb RAM, 486 processor.
 Microsoft Windows 95, 20Mb RAM, 486 processor.
 Microsoft Windows NT, 32Mb RAM, 486 processor.

ESSENTIAL READING

Notes on, an introduction to network modelling, network simulation, and Comnet III.
Comnet III Models (Use the on-line help for the models used in this exercise).
Comnet III reports (Use the on-line help for details on the reports).

FURTHER READING

Comnet III on-line help (Note: The on-line help is the Documentation supplied with the software).
Web site for CACI (www.caciasl.com).

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 2
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
REQUIREMENTS

 Before considering the network modelling package you should have carried out the necessary
preparation (i.e. pre-reading and note taking).
 You should take notes before, during and perhaps after the practical exercise.
 Before, during, and possibly after the modelling and simulation process you should address the
questions given below.

QUESTIONS FOR THE NETWORK SIMULATION EXERCISE

1. Calculate (using pen and paper), the time required to transfer a file of 60 bytes across links with
speeds of; 100bps, 300bps, 1200bps, 2400bps, 9600bps, 19200bps.
2. Describe how you would expect the system to deal with a source which is generating data at a
higher bit rate than the link bit rate ?
3. Is the transfer time greater, smaller, or the same for the scenario of 1 Byte values vs normal packet
sizes ? Explain the results.
4. Discuss any significant simplifications that have been applied in creating the model of the ‘real’
system.
5. How might the performance of the ‘real system’ differ from the simulated system results ?
6. How do your calculated results (from 1.) compare with the simulation results ?
7. Is there an optimum link speed ? If so, state the link speed and explain why it is the optimum ?

SCENARIO
A water authority uses remote computers with detectors attached to them. The remote computers take
readings from the detectors, which monitor the condition of the water at various stages in a water
purification process. Remote computers are accessed from a local computer, and data is downloaded.
If the data is processed by the remote machine it is transferred across the link one byte at a time, at
one second intervals. If the data is processed by the local machine, the data is transferred in varying
packets sizes at one second intervals. The packet sizes are random, but are known to be normally
distributed, with an average size of 10 bytes, and a Standard Deviation of 5. The company is
concerned that the link may be insufficient to handle the varying packet sizes which are produced
when the raw data is transmitted for processing by the local machine. A model of the network is
required for analysis, in order to determine whether the 110bps link is sufficient, and whether any
other performance factors should be taken into account.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 3
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
USING THE COMNET III SOFTWARE
INTRODUCTION

Comnet III is a comprehensive performance analysis tool for computer and communication networks.
Based on a description of a network, its control algorithms and workload, Comnet III simulates the
operation of a network and provides network performance. No programming is required, as network
descriptions are created graphically. A description of the package and procedures for the construction
of network models follows. The Comnet III interface is shown in figure1.

Figure1. The Comnet III Interface under Windows 95.

THE TOOL BAR.

The Tool Bar is used in a similar manner to any windows based software. It allows for rapid access to
commonly used menu items. Figure 2 below, shows the function of each of the tools.

Figure 2. An annotated diagram of the Comnet III toolbar.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 4
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
THE PALETTE.

The palette is shown in Figure 3. It is used to select the required network modelling objects. To select
a modelling object the ‘select objects’ tool must first be indented (clicking on the tool will cause it to
toggle on/off); then click on the required object in the palette and drag it to the required position on
the screen. Once the object is in position, click the left mouse button to fix the position. To create a
connection (or association) between two objects, select the diagonal arc tool (or Horiz/Vert tool), click
on the first object then click on the second object. A connection arc will be displayed between the two
objects.

You will only be using the objects listed below for this exercise. Note their position in the palette.

1. The diagonal arc (also referred to as a connection tool).

2. The processing node.
3. The point-to-point Link.
4. The message source (also referred to as a traffic source).

Figure 3. Annotated diagram of the Comnet III palette.

OVERVIEW OF THE PROCESS FOR BUILDING A NETWORK MODEL.

A model is built and executed using the following steps:

 Nodes, links and traffic source elements are selected from the palette and dragged into position on
the screen.
 These elements are connected (using a connection tool) to define their interrelationships.
 The user double clicks on either nodes, links or traffic sources. A dialogue box with all adjustable
parameters appears and the user specifies the parameters for the particular item.
 Network operation and protocol parameters and the required reporting statistics are set on
additional dialogue boxes accessed through the menu bar.
 The model is verified and executed, after which the results are presented in various reports.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 5
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
PROCEDURE FOR CREATING AND RUNNING A NETWORK MODEL

You will now build and simulate a model of a network.. The network model is based on the network
described in the scenario section of this exercise. The network includes one remote station which
sends the message data, and one local station. The stations will be connected using a 110bps point-to-
point link. The complete network model is shown in figure 4 below.

Figure 4. Diagram of the complete Comnet III network model.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 6
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
PROCEDURE FOR CONSTRUCTING AND SIMULATING THE NETWORK MODEL.

Stage 1 - Loading the Comnet III software.

 Locate and load Comnet III on the machine (double click the ComnetIII Icon and wait!!).
 Click on the software license key followed by OK.
 Click Cancel for the readme file.
 You should be presented with the screen shown in Figure 1.

Stage 2 - Creating the model (Refer to figure 4).

 Select a processing node (Local computer) and place it in the required position on the screen.
 Select a second processing node (remote computer) and place it in the required position on the
screen.
 Select the point-to-point link and place it between the processing nodes.
 Select a message source and place this next to the remote processing node.
 Select the diagonal arc link connection tool and create associations between the objects.

The network objects and associations should look similar to those in figure 4.

Stage 3 - Naming the objects and changing the parameters.

You will now give each of the objects you have created, a more meaningful name, in addition to
changing some of the parameters for the objects.

The Message Object

Change the name of the message icon as follows:
 Double click the message icon.
 You should now be presented with a dialogue box.
 Change the Name field to ‘MsgRemote’.
 Change the Interarrival time to one second by clicking the down arrow and selecting 1.0.
 Change the Message size (Prob distrib) to one unit by clicking the down arrow and selecting 1.0.
 Note that the size of a message ‘msg size units’ unit, is Bytes.

The Link Object

Change the name of the link, and link parameters as follows:
 Double click the link icon.
 Change the name to ‘Link110bps’.
 Click on the double dots, next to the Parameters box.
 Double click on DEFAULT (A dialogue box should appear).
 Change the bandwidth/circuit [kbps] [from node X] to 0.110.
 Change the bandwidth/circuit [kbps] [from node Y] to 0.110.
 Click on ‘Done’, and exit the dialogue boxes.

The Processing Node Objects

 Change the names of the processing nodes by double clicking the processing node icons and
naming them Remote and Local as shown in figure 4.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 7
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
Stage 4 - Selecting the reports (Use On-line Help for further information on the available reports and
report details).

You will now specify the statistics you would like Comnet III to produce.
 Click the Report menu followed by Select Reports... (You should now see the reports dialogue
box).
 Double click on the Links[+] line. (This will display a list of the reports which may be generated
for the link).
 Click on ‘Set All’ to turn on the report generation statistics for ALL the link parameters. (You may
wish to select further items for report generation at your own discretion).
 Click ‘Done’ and exit the dialogue boxes.

Stage 5 - Running the simulation.

You will now examine the run-time parameters, and then run the simulation.

 Click on the Simulate menu followed by Run Parameters. (A dialogue box should appear).
 Note the run time (Replication Length) is 60 Seconds.
 Click on cancel to close the dialogue box.
 Click on the green ‘traffic light’ on the speedbar to run the simulation.

After the model has been verified, you will be asked to save the model. Save the model to drive C:, in
the folder Comnet, and use the name expt4.c3. Note: If the Comnet folder does not exist, you will be
required to create one.

Stage 6 - Reading the reports.

 Click on the Report menu followed by Browse Reports.

 Select the expt4 folder.
 Double click on stat1.rpt. (The results are now displayed).

You can now print the reports in the usual manner (use the file menu followed by print).

FURTHER SIMULATIONS

1. Change the message traffic size from 1.0 units in size to a Normal distribution with an average
size of 10 units, a Standard Deviation of 5, using random number generation stream 2. (i.e. Select
Nor(10,5,2)).
2. To assist in determining solutions to the questions posed, try changing parameters for the
simulation (e.g. link speed), and re-running the simulation.

ON COMPLETION OF THE EXERCISE, PLEASE REMOVE THE COMNET FOLDER ON DRIVE

C: (click the right hand mouse button on the Comnet folder, and select delete from the menu).

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 8
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
NOTES: AN INTRODUCTION TO NETWORK MODELLING,
NETWORK SIMULATION, AND COMNET III.
Why model and simulate networks?

There is an increasing emphasis on computing and communication networks in the day-to-day

operation of all types of organizations. As these networks become larger and more complex the design
and management of the system becomes an even more challenging task. “Back of the envelope”
calculations are not a reasonable basis for the design validation of a multimillion dollar network!
Elaborate spreadsheets won't do the job because of the random or ‘stochastic’ nature of network traffic
and the complexity of the total system. New technologies are always being introduced. New
applications of communication and computing networks are constantly being explored. How do
organization decide which combinations of technology and applications are right for them? As needs
grow the needed investment and operating costs increase. As the organisation places more
dependence on its network to support its critical business operations, the risk of failure caused by poor
network performance or availability may have serious repercussions. How can these alternatives be
assessed at an early stage and the design of the network adapted to minimise risk?

System designers and network planners are increasingly looking for support tools to help them decide
on the best design. Results from tools which allow performance prediction through simulation may be
used to help support their cases to clients and management. For example, a model may be used to
assess different design alternatives, or different operational policies. The analyst can explore the
behaviour of a proposed system without actually building it. It would be expensive, impractical, or
even impossible to build 3 alternatives, pick the best, and throw the other two away. Alternatively,
planned modifications to an existing system can be pre-tested through simulation without disturbing
the actual network. A bank cannot take down its mainframe computer centre during peak hours to test
its recovery time performance.

What will Comnet III do?

Comnet III lets you analyze and predict the performance of networks ranging from simple LAN’s to
Complex enterprise wide systems. Comnet III supports a library of objects that closely model the
objects used in real networks, with one Comnet III object representing one or more real world objects.
The default Comnet III object parameters may be adjusted to match the real-world object.
The object-oriented framework of Comnet III provides the flexibility to try unlimited ‘what if’
scenarios. During simulation, you will be supported by an animated picture of the network
configuration you have selected - no programming is required.
COMNET III can be used to model both Wide Area Networks (WAN’s) and Local Area Networks
(LAN’s). COMNET III models may contain both types of facilities in one integrated model.
COMNET III can also provide detailed modelling of network node logic. A node's computers, their I/0
subsystems, their databases and the applications which run on the computers can all be modelled.

The network modelling approach used in COMNET III is designed to accommodate a wide variety of
network topologies and routing algorithms. These include:

 LAN, WAN & Internetworking systems

 Circuit, message and packet switching networks

 Connection-oriented and connectionless traffic

 Static, adaptive and user-defined algorithms

A significant feature of COMNET III is the ability to abstract portions of a network model and treat
them as modular components. This capability follows from the object-oriented design of COMNET
III. This facility also allows the user to build a library of network components which can be "plugged
in" and swapped at will.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 9
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
What could you use network simulation for ?

Typical network simulation applications include:

 Peak Loading Studies

Generally a network is subject to heavy levels of traffic at particular times of the day, week,
month or year. If the network design can cope with this level of traffic then it can cope with
the workload during other periods. A typical use of network simulation is therefore to model
these peak loading periods to gain an understanding of the stress points in the network.

 Network sizing at the design stage

When designing a new network some provision for growth has to be allowed for. Network
simulation can be used to assess that the design meets current traffic levels, and it can be
used to see what room there is in the design for system growth.

 Resilience & contingency planning

It is often important to know that a network design has sufficient resilience to offer a
reasonable level of performance in various failure scenarios. The nodes and link components
in a network simulation model can be failed and recovered at various times in the simulation
to test various contingencies that are not testable in the real system.

 Introduction of new users/applications

New users and/or applications will typically add more load onto the network. It is useful to
try and predict their impact before their introduction so that potential bottlenecks can be
identified and resolved before a major problem appears.

 Evaluating performance improvement options

Many networks have year on year traffic growth. This results in deteriorating network
performance until the network is upgraded in some way. The various options for upgrading
can be investigated in using network simulation as part of a cost vs benefit study.

 Evaluating grade of service contracts

It is increasingly common practice for service level contracts to be negotiated between the
network user and the network provider, even when they are part of the same organisation.
Network simulation can be used to analyse the performance service levels that can be
attained during contract negotiation, and to predict potential problem areas as usage patterns
of network components change over time.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 10
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
Network components and Comnet III models

A network is taken to mean an arbitrary interconnection of computing and communication devices for
voice, data, video, or other types of network traffic. These may include terminals, workstations,
servers, network interface adapters, connection media (i.e. Ethernet cable, twisted pair), repeaters,
bridges, gateways, routers, pads, front end processors, hubs, packet switches, PTT exchange
equipment, pbx's, handsets, leased lines, satellite links, etc. An organisation will have it’s own
particular network design which may include some or all of the equipment types listed above and
possibly others that we not listed.

The goal of COMNET III is to provide the capability to include any network equipment type in the
simulation. The user interface provides flexible interconnection of different devices so that you can
describe your network to the system. COMNET III does not provide a list of every possible device ever
built and used in a network as this would be an impossible task. Rather, it uses generic building
blocks which can be parameterized to represent the devices you want to model. For instance, you may
be a publisher and want to model a LAN system which has several print servers connected to it.
Printing large image files is the main bottleneck in the system. COMNET III does not have a print
server object, but it does have a Computer & Communications Node which can be used for the
purpose of receiving and processing printjobs. Consequently, when you use COMNET III you have to
look at your real network and make mappings of the devices and functions you have in your network
to the COMNET III constructs, or add new components to the COMNET III model.

A computing network is a system like a Local Area Network (LAN) or a computer centre. There is a
population of users connected to it and they demand applications to be run either on their local
workstations, remotely on a server, or on a mainframe.

A telecommunication network is generally a bearer system like a backbone network or a public service
network such as an X.25 system. It may be private to the organisation, or it may carry third party
traffic as in a PTT or commercial telecommunications system. As with a computing network, there are
users connected to the network but, from the point of view of the network operator, there is little
knowledge of the end systems that are being serviced. For instance, an X.25 service provider does not
know whether you are using a PC or a workstation or what software application you are using to send
messages over the network- nor does the provider know what the recipient will do with any received
information. What the service provider sees is simply a demand for traffic flow from an origin to a
destination. The provider wants to fulfil this requirement as quickly and reliably as possible. Wide
Area and Metropolitan Area networks fall into this category.

There is a rapid growth in internetworking. This is where computing networks are interconnected
over communication networks (WAN or MAN) to provide access from one computing network to
another. A major concern in the design of internetwork systems is the adequacy of data transmission
rates offered by the telecommunications network. Computer to computer traffic normally expects to
see a high speed LAN.

Voice networks can be also be modelled with Comnet III. Trunk capacity, routing and peak loading
issues can be investigated.

Choosing the Correct Level of Detail

Whichever type of network you are modelling, you have to pick the right level of detail in the model
to answer the questions that are important to you. This is sometimes referred to as the granularity of
the model. Think carefully about this aspect of modelling as it will greatly influence the degree of
success that you have with simulation.

Not enough detail and you may miss some important aspect of the system's behaviour. Too much
detail and you will end up with a model which is larger than needed and which takes longer to run
than necessary for each experiment.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 11
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
A Backbone Network modelling example.

Consider the backbone network for a bank. The banks application development department is about to
introduce a new application in all of the banks branches. You know the size and frequency of
transactions to be posted over the network, and you have estimates of the number of users and their
geographical locations. Given this information, you would like to know whether current response
times will be degraded when the new application comes on line. If this is the case, you must identify
the cause of the bottleneck. In this case it could be inadequate leased line capacity or inadequate
capacity in the network routers. It is important to know in advance whether potential bottlenecks can
be avoided by acquiring additional leased line capacity or by upgrading routers.

A model of this situation can be constructed using the COMNET III nodes to represent the pads the
branch offices are connected to, the backbone switches/routers, and the mainframe computer centre.
These are connected with point-to-point links to represent the leased lines in the system. The technical
characteristics of the pads and routers are basically their throughput rate (switching times), while the
point to point links are defined in terms of their speed. On top of this network you define a traffic load
which originates at the pads to represent the user demands on the system. This traffic is defined as
messages which carry the transaction data between the pads and the computer centre and is specified
by how often messages occur and how big they are.

Building a model like this allows you to determine if leased line capacity or router speeds will cause a
bottleneck in your system. The model contains sufficient details about their performance
characteristics to reasonably model them, and also sufficient information about the traffic load.

As an alternative to the model building approach you could have done a simple spreadsheet or "back-
of-the-envelope" model by adding the average transaction rates from all of the branches and
comparing this with the rating of the equipment in the computer centre. For example, 50 transactions
per second from users with 100 transactions per second capability on the computer centre. The
problem with this approach, however, is that it does not allow for the bursty nature of traffic. In real
systems you would not have a transaction every 20 ms exactly. If there is congestion, a spreadsheet
does not tell you where it will be or allow you to investigate solutions. In other words, the real world
is not known for its ability to present you with events that occur at evenly spaced intervals at the
known average rate. The variability of real world situations can only be handled with a true discrete
event simulation model of the type provided by COMNET III.

Of course, you could also build a COMNET III model with more detail than this. You could assert that
modelling down to the branch level and aggregating traffic there is not sufficient, and that you want to
model every user connection to the system. This will increase the size of your model with little or no
increase in accuracy of the results. You will have much more information, but the additional
information will not contribute to your knowledge about leased line or router adequacies. You would
be faced with the task of collecting and entering more data into the system, you would have longer run
times and you would have larger report files to examine because of all the devices you have modelled.
Your model would be larger than it needs to be to provide the answers to the particular questions
regarding line capacity and router speed.

If you added the requirement to examine the effect of the proposed change on individual users, then a
more detailed model is justified, but it is not necessary to explicitly model every single user in the
system. The approach which is often used in this situation is to model one or several users connections
explicitly while aggregating the remainder of the users' connections. This gives performance estimates
for the single user without overloading the model (and yourself) with detail.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 12
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
A Local Area Network modelling example.

At the other end of the scale from backbone Wide Area Networks, consider modelling the performance
of a local area network (LAN). LANs are characterised by a high speed transmission medium (e.g.,
10 MB/sec for Ethernet) interconnecting workstations, servers, printers, gateways etc. If you are
running a database application over a LAN, it is often insufficient to model just to an arrival stream of
inquiries with a corresponding stream of replies. While this will probably give a reasonable
prediction of LAN utilisation and transmission queuing delays over the Ethernet, it will not tell you
what aspects of workstation or server operation are contributing to delay time for inquiries. In this
case you will have to model the end systems, such as the database server, in more detail in order to see
their affects on the system.

Modelling more detail on end systems in COMNET III is achieved by modelling more of the
hardware characteristics of the end systems, together with the software load that is being placed on
them. It is generally the case that poor server performance is caused by many users demanding
concurrent processing or data retrieval on the server. If desired, COMNET III can model server
performance in considerable detail. The speed of the processor (i.e., CPU cycle time) and disk access
times can be specified together with the sequence of software actions that the server undertakes in
response to any particular demand. For each of the software actions in a sequence, the load that it
places on the system (in terms of CPU cycles) can be specified.

Summary: Level of Detail

You have to choose the appropriate level of detail for the system you are trying to model and the
performance questions you are asking. COMNET III is capable of modelling on many levels, from
modelling a specific subroutine running on a specific computer in a world-wide network, to
aggregating the throughput capacity of the same network as a number of transactions per second.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 13
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.
Consideration of accuracy

Comnet III is designed to accurately estimate the performance characteristics of computing and
communication networks. Estimating means that the network under study is described to COMNET
III via data. COMNET III then executes a dynamic simulation of the network which builds a
computer representation of the network and routes simulated traffic over it. Reports are produced on
the measured performance of the different model elements and overall network characteristics, and
are presented as the estimations of network performance.

This data, which is entered via a graphical user interface, describes:

1. The topology of the network: nodes, computer centres, connectivity, etc.

2. The workload placed on the network. This includes the applications that run on end systems and
the traffic to be delivered across the network. The frequency and size of different tasks may be
described statistically.

3. The protocols or rules for scheduling applications and routing traffic.

The reports produced are an estimate of the expected performance of the real network. Their accuracy
is dependent on the data that has been entered to describe the network. One of the major questions is
how accurate is the data and consequently how accurate are the estimates of performance.

Another factor which determines accuracy is the run-length or amount of simulation time the model is
run. The length of the run determines how many random events are used to represent the statistically
generated traffic. For instance, you may specify that file transfers are to be modelled, and that the file
sizes are randomly picked between 10KB and 50KB. If you only run the model long enough to
represent 5 file transfers then the file sizes might be 37KB, 21 KB , 17KB 11KB, 31KB which gives
an average file transfer of 23.4KB versus an expected average of 30KB. However, if you run the
model longer and obtain results over 1,000 file transfers, then the average file transfer size will
converge on the expected average.

With run length in mind, the accuracy of the results of a simulation are normally quantified with a
variation and a statistical confidence estimate. For instance, file transfers are completed with an
average of delay of 10.5 seconds with an observed standard deviation of 2.3 seconds and with a 95%
confidence of statistical correctness.

COMNET III can run multiple, independent replications of the simulation and generate mean,
maximum, minimum and standard deviations, as well as plots and histograms of system performance.

Once you have built a model which produces accurate estimates of the performance of your network,
you can then use the model for a variety of “what if” experiments.

Exercise 4. An introduction to network modelling and network simulation using Comnet III. 14
Developed by Iain Hughes. 12/13/2019 12/13/2019 Document comintro.doc.