Coursework Assessment 2019-2020 Network Modelling and Performance Evaluation – Network links upgrade

Coursework Assessment      2019-2020

Network Modelling and Performance Evaluation – Network links upgrade 

 

Course: COMP1664  Network Technology	Contribution:  100 % of course
Network Technology – Term 2 – MAC	PDF file required
Greenwich Course Leader:	Due date: 20th April 2020
This coursework will be marked anonymously YOU MUST NOT PUT ANY INDICATION OF YOUR IDENTITY IN YOUR SUBMISSION
This coursework should take an average student who is up-to-date with tutorial work approximately 50 hours
Learning Outcomes: Compare and critique network types, protocol architectures, hardware and software components, and evaluate their pros and cons. Connect and integrate various network technologies to achieve required configurations and end-to-end systems. Evaluate behavioural aspects of networks and the ways in which network traffic and network performance interplay. Design, build and use models of networks and evaluate their performance.

 

 

Objectives:

·             To enable students to investigate network technology through simulation.

·             To enable students to investigate the effects of various network traffic conditions on network performance.

·             To enable students to gain experience of carrying out an experiment, to collect and process experimental data, and to present and analyze results with clarity and depth.

·             To enable students to appreciate the role and value of simulation tools when designing and analysing networks.

·             To allow students to gain practical experience of using a network modelling tool.

 

Work Mode:

This task is to be completed individually.

 

The Task:

The emphasis of this task is to design and use a model of a network as the basis for an investigation into the factors that affect the network’s performance (using throughput and delay as performance metrics). It is necessary to build a realistic model of a network in order to perform reliable experiments and achieve meaningful results.

This coursework task is divided into three parts, plus a conclusion. The Interim submission requires parts A and B, the Final submission requires parts A, B and C, plus a conclusion. For each submission you should upload a single report.

Part A – Designing a network and modelling it

1.     Design a network for a medium-sized company, or an organisation such as a hospital or airport, located at a single campus. You can base the design on the requirements of an existing company or organisation, or you can invent one. In either case care must be taken to ensure the design is realistic. You need to clearly identify the organisation’s line of business and the data and communication needs they have, and thus determine the consequences this has on the network requirements.

Some constraints are provided to guide you:

–      The network must be designed taking into account the type of business or organisation you have chosen, and thus must be appropriate for their network traffic requirements.

–      The network must be divided into at least three main sub-networks; each representing a department or a group of departments within the organisation.

–      The sub-networks must be interconnected with a network core containing at least three routers.

Apart from these constraints, you have freedom over the actual design, but it must suit the needs of your chosen organisation and you need to justify this aspect.

2.     Critically discuss your design, explaining the topology and technologies used. Provide justification in terms of the company you have chosen, and their specific networking needs.

 

3.     Create and test a model for your design (for part A only the topology aspect of the model is expected).

 

Part B Network Traffic Modelling

You are required to evaluate the performance of your network under realistic conditions. Workload is constantly fluctuating in real networks and can vary significantly; therefore you should test the network under a range of different workload types and intensities.

This step, see below, sets up your base workload model on which you perform the majority of experiments (see section C). However, it is often interesting to test the network under significantly higher workloads than are expected to occur – to ‘stress-test’ your network design. To do this you could try doubling or quadrupling the traffic intensity in your base workload model.

Devise a realistic network traffic model for your organisation:

i.        Determine at least three staff profiles, e.g. by department, so there might be profiles defined as (e.g.)  {sales, administration, … , … } depending on your organisation scenario. Identify the roles of the staff in these profiles (what type of work do they do).

ii.       Determine, with justification, the types of application traffic e.g. {Email, FTP, HTTP, Database, … } generated by each staff profile. It is important to choose a realistic set of network application types that would be used by the staff of the organisation you have chosen. Three or four applications for each profile is adequate; choose the ones that are most relevant to the type of work the organisation performs. For each type of application, determine the intensity of network usage that arises (in terms of the size of messages generated and the frequency of messages generated). There are default settings for ‘heavy’ or ‘light’ usage, but you need to determine and justify the appropriate values for your specific scenario.

iii.     Configure your model to reflect this network traffic model. This requires that you set the interarrival times and message sizes appropriately. Run your model to make sure the settings are correct and that network traffic is actually being generated in your network and that the levels of traffic are realistic in the context of the application scenarios you have devised.

 

Part C – Network Performance investigation

The emphasis of this stage of the work is to understand the relationship between the network’s configuration and its performance. You need to perform a sufficient number of separate experiments in order to be able to see the actual effects on performance of specific network configurations or changes in traffic conditions.

1.     Investigate the ways in which changes in the link performance (in terms of bandwidth) affect the network performance, using a variety of traffic intensities.

Specific details of the methodology to be followed:

Starting with all links in the network having 100Mb/s bandwidth, perform experiments to measure the delay and throughput at suitably representative selected points in the network. We can denote this experiment 100_100 (i.e. the access bandwidth is 100Mb/s and the core bandwidth is 100Mb/s)

Increase the bandwidth of the core layer links to 1Gb/s (the access layer links remain at 100Mb/s), and repeat the experiments. We can denote this experiment 100_1000.

Increase the bandwidth of the core layer links to 10Gb/s (the access layer links remain at 100Mb/s), and repeat the experiments. We can denote this experiment 100_10000.

Set the bandwidth of all links to 1Gb/s (the access layer links and the core layer links), and repeat the experiments. We can denote this experiment 1000_1000.

Increase the bandwidth of the core layer links to 10Gb/s (the access layer links remain at 1Gb/s), and repeat the experiments. We can denote this experiment 1000_10000.

You will now have five sets of experiment results for each of the two performance metrics.

The experiment denotation {100_100, 100_1000, 100_10000, 1000_1000, 1000_10000}, as explained above, will form the X axis (independent axis) of your graphs, similar to the example graph, figure 1, but replacing the ‘number of active users’ as was used for that example.

The performance metrics used will be Throughput and Delay. Hence you need to provide two sets of results, one for each of these performance metrics.

Collect the results and present them in a clear format that enables comparison of the effect on performance of each workload profile. There should be two graphs (one for each performance metric). Export the data from your model into Excel, and use Excel graphs to plot the results.

In both graphs, the Y axis (dependent axis) should be the measure of performance against one of the performance metrics. The graph format should resemble that shown in figure 1, but with the X axis labels changed to the link configuration denotations {100_100 … 1000_10000}.

 

Figure 1. An example of the expected graph format. Note that it has meaningful axis titles and main title. Also note that the trace lines are thin so the actual value can be accurately read at any point, and also note that the trace lines each have a different drawing style, so the graph can still be understood even if printed in black and white.

2.     Analyse and critically evaluate your results. In your report you should explain the experiments that you have carried out, providing configuration details etc. There should be a general introduction to the results section, providing overall explanation of the experimental methodology used and the specific experiments carried out.

For each experiment there should be a brief explanation of the experiment itself, the actual results (in a table, and also in graphical form), and there should be a paragraph explaining the results in each particular graph. Make sure you discuss the data values and the metrics used in a clear and unambiguous way. For example, if describing figure 1 above, you would need to say where and how the delay was measured (i.e. was it average end-to-end delay? worst-case end-to-end delay? delay on a specific link? or what? – this of course depends on which parameters you have selected to monitor, when running the simulation). Analyse and critically discuss each set of results, explaining the particular shape of the curves. Explain the causes (in the network design, traffic configuration etc.) which have given rise to these specific results, and the significance of any trends or patterns. In addition, at the end of the section provide some overall discussion of the results as a whole, and what they indicate about the performance of the network and its suitability for your organisation scenario.

Deliverables:

An optional interim report. You are strongly advised to submit this, by the advertised deadline, in order to receive valuable feedback. However there is no penalty for non-submission of the interim report.

Only parts A and B are expected. You need to show some progress towards these parts but the work is not expected to be complete at this stage.

           

A mandatory final report:

You should submit a single, well-structured report consisting of 15-20 sides of A4 which must include three sections clearly marked parts A, B and C, as well as a conclusion of 200 – 400 words.

 

Deliverables for part A:

·       Detailed and clear documentation for the network design. Include critical discussion of your design, justifying the topology and technologies used. Provide justification in terms of the organisation you have chosen, and their specific networking needs.

·       Topology screenshots of your model, with brief explanatory notes.

Deliverables for part B:

·       Detailed and clear documentation for the network traffic model, including types and numbers of users, applications and traffic generation configuration. Include justification of the network traffic model, in the context of the type of organisation you have chosen. You should explain any assumptions and justify your decisions.

Deliverables for part C:

·       Results from the simulation study (in tabular form with Excel graphs). Provide a detailed explanation of the meaning and significance of each graph.

Conclusion:

·       A clear statement of what has been achieved, what you have learned, and what could be improved.

 

Grading criteria / Allocation of marks:

The allocation of marks will reflect the quality of the work produced.

 

·     To achieve a first class mark a student will need to produce a professional quality report that is technically correct, well structured and includes substantial evaluation, critical appraisal of results and justification of design decisions.

·     A technically correct report with a good attempt at evaluation and justification will be awarded an upper second class mark.

·     A technically correct report with limited evaluation or limited justification will be awarded a lower second mark.

·     A report that is incomplete, technically incorrect, lacking technical detail or lacking evaluation or justification will get a lower second or third or fail grade depending on content.

 

Marking scheme:

Part A
Detailed and clear documentation for the network design with justification	20%
Topology screenshots of model, with brief explanatory notes	10%
Part B
Detailed and clear documentation for the network traffic model with justification	30%
Part C
Results from the simulation study (in tabular form with Excel graphs), with accompanying explanation	30%
Conclusion (200 – 400 words)
Statement of what has been achieved, what you have learned, and what could be improved	10%

 

IMPORTANT NOTES:

Failure to include one of the sections identified in the marking scheme (above) will result in forfeiture of the marks allocated to the missed section.

Template:

Coursework solution templates (for MS Word) are provided at the course website. Please download the appropriate template (interim or final) and use it as a framework for your report. The templates have been designed to help you provide a high-quality report.