COMP90007 - Internet Technologies Network Analysis Project - Project 1 - Solved

1         Introduction
This project forms 10% of your final mark. The key output of this project is a report which has to follow a certain format (refer to Section 5 of this document). The project is about measuring bandwidth, delay, and jitter in networks. These tasks will be similar to those you have performed in the laboratory sessions held in the tutorials. It is recommended that you perform these tasks in a consistent networking environment to reduce the variance in your report.

Important Note: As evidence of your work, when you run the following commands, please remember to take screenshots of the results obtained and place it in the appendix of your report. Reports failing to do so will be penalized. All plots, however, should be placed in the main body of your report where you will explain the observations being derived.

2         Measuring the hop count
In this section, we will be observing the number of intermediate hosts in the route taken to communicate with a remote server and its relation to the physical geographical distance.

To count the number of hops taken to reach a destination host, the command tracert will be used (or its corresponding equivalent, depending on your operating system). This utility should be pre-installed on your operating system.

The utility can be invoked by launching a command line terminal and typing in the command.

An example output of the traceroute command (on OS X) and tracert command (on Windows) is as follows:

$ traceroute -nw1 cis.unimelb.edu.au traceroute to cis.unimelb.edu.au (128.250.37.164), 64 hops max, 52 byte packets

1            10.0.0.254 533.676 ms 1.063 ms 0.940 ms

2            58.96.2.205 27.872 ms 28.137 ms 28.293 ms

3            58.96.2.129 28.647 ms 28.577 ms 28.085 ms

4            218.100.78.33 28.299 ms 28.469 ms 28.332 ms

5            202.158.200.9 29.626 ms 28.871 ms 29.841 ms

6            202.158.210.26 31.320 ms 28.722 ms 29.135 ms

7            202.158.200.250 29.668 ms 29.096 ms 28.660 ms

8            * * *

9            * * *

10         * * *

11         128.250.37.130 957.521 ms 33.475 ms 29.891 ms

12         128.250.37.164 29.940 ms 29.260 ms 30.020 ms

In this section of the project, you are interested in the number of hops it takes to reach the destination server. In the example above, the number of hops to reach cis.unimelb.edu.au is 12.

Based on the number of measurements you will be taking, there are some useful command line parameters you may wish to take advantage of, to speed up the time it takes to gather results. The help documentation for the traceroute utility can be accessed by running man traceroute or tracert /? on Windows. You may also wish to investigate shell scripting to automate the collection of results, but this is not required for the project. Any scripts you do choose to write, however, must be included in the Appendix.


2.1 What do the command line parameters -n w 1 (equivalently -d -w 1 on Windows) mean in the example given above and what is the importance in using them?

2.2 Determine the hop count for the following hosts given in Table 1. It is recommended that students find one more public iperf server other than the ones listed here to gather their results.

Table 1: List of public iperf hosts

Host
Location
iperf.he.net
USA
bouygues.testdebit.info
France
iperf.comneonext.de
Germany
ikoula.testdebit.info
France
st2.nn.ertelecom.ru
Russia
iperf.biznetnetworks.com
Indonesia
iperf.scottlinux.com
USA
speedtest.serverius.net
Netherlands
iperf.volia.net
Ukraine
Determine the approximate geographical distance for the above hosts and plot the hop count versus the approximate geographical distance from Melbourne. Do you observe a correlation? Why? Why not? Explain your rationale with respect to networking concepts.

You may use any scientific computing package or spreadsheet software to do your plotting, for example, Microsoft Excel.

For finding out the physical grographical distance you may use any tool or application available online, for example, you may use a combination of: https://db-ip.com and https://www.freemaptools.com/how-far-is-it-between.htm or https://www.site24x7.com/findwebsite-location.html and https://www.distancecalculator.net/ or anything of your choice. However, do make sure to document it and provide the appropriate reference to that application/ tool/ software used.

Note: The servers listed in Table 1 are public servers and are not maintained by the University of Melbourne, hence they are likely to go down at any point in time. Based on past experience, it would be advisable to conduct your tests on these as soon as possible (ASAP) rather than leave it till the end of the time frame because there is a high probability that these servers might not be available and this cannot be used as an excuse for a late submission. If these servers stop responding then please visit the link: https://iperf.fr/iperfservers.php and find your own servers (anything that is responding) or feel free to find any public iperf server from the internet, including deploying your own iperf server.

3         Measuring delay and jitter
In this section, you will be measuring the delay and jitter of the hosts listed in Table 1, located in different geographical locations.

We will be using the ping utility, to measure the round-trip delay of packets. The ping utility should be pre-installed on all major operating systems. The standard deviation of the round-trip delay time will be taken as the value for jitter for this project.

The standard deviation measures the variation in a set of data. It is defined as the square root of the variance and is expressed as follows:

 

where x¯ is the mean of the set of data. Details of this simple statistical measure can be found in many sites online if you do not remember this from high school years.

A sample output of the ping utility is shown below, but this output will vary depending on your operating system.

$ ping unimelb.edu.au

PING unimelb.edu.au (172.22.44.10): 56 data bytes

64 bytes from 172.22.44.10: icmp_seq=0 ttl=124 time=3.364 ms

64 bytes from 172.22.44.10: icmp_seq=1 ttl=124 time=3.416 ms

64 bytes from 172.22.44.10: icmp_seq=2 ttl=124 time=3.730 ms

ˆC

--- unimelb.edu.au ping statistics ---

3 packets transmitted, 3 packets received, 0.0% packet loss round-trip min/avg/max/stddev = 3.364/3.503/3.730/0.162 ms

In the output above, various statistics, including the mean and standard deviation, were calculated for you by the utility. It is expected that you record all the values of delay and perform your own calculations to find the mean values and their standard deviation, to confirm the result.


3.1 Measure the round-trip delay for the following hosts. Make three delay measurements (run this command 3 times - not 1 command gathering 9-10 rows of ICMP responses) of each host and find the average round-trip delay and jitter by calculating the standard deviation, for each host in Table 1.

For each of the above hosts, plot the average round-trip delay versus the approximate physical geographical distance to the server. Do the same with the jitter (i.e. jitter vs geo distance).

3.2 From the two plots above, do you observe any correlation between delay and jitter as a function of distance? Why? Why not? Explain your results with reference to the network environment in which you were collecting your results (this includes metrics like your download/ upload speed, users sharing the network, load on network through other apps, etc) and how does your networking environment influence your results obtained (provide examples)?

4         Measuring the bandwidth-delay product
In this section, we will be measuring the bandwidth of different hosts in order to determine the bandwidth-delay product, using the results from the previous sections.

The utility that will be used to perform bandwidth measurements will be the iperf utility. This command line utility is available for download for all operating systems from https://iperf.fr. Alternatively you may choose to use the package manager for your relevant operating system.

There are two modes of operation in iperf. The server mode will host a server which will listen to incoming requests from a client. An iperf instance running in client mode will connect to the server, and packets will be exchanged and timed between the two hosts to calculate the bandwidth. In this project, we will be running iperf in client mode.

A sample output of iperf in client mode is shown below, noting the -c flag to designate operating in client mode.

$ iperf -c iperf.eenet.ee

------------------------------------------------------------

Client connecting to iperf.eenet.ee, TCP port 5001

TCP window size: 129 KByte (default)

------------------------------------------------------------

[ 5] local 10.0.0.5 port 51878 connected with 193.40.100.7 port 5001

[ ID] Interval                                 Transfer                 Bandwidth

[ 5] 0.0-106.7 sec                                      128 KBytes 9.83 Kbits/sec

(Note: speedtest.serverius.net server) For speedtest.serverius.net in Table 1, we may need to use the port 5002, so the command will be: ”iperf3 -c speedtest.serverius.net -p 5002” for this host. Also, some iperf servers respond to iperf2, rest to iperf3 so you might want to use both iperf2 and iperf3 to verify if the server is responsive.



4.1 What does the bandwidth-delay product tell us about the networks? Collect three set of measurements (run this command 3 times) measuring the bandwidth of the public iperf hosts in Table 1 and find the mean bandwidth for each host.

4.2 Take the mean bandwidth and calculate the bandwidth-delay product in kilobits. You may use the mean round-trip delay time from your ping experiments to use as the delay time. Plot a bar chart for each host showing your results. You may wish to use a logarithmic scale, if appropriate.

Explain your results making reference to your networking environment in which you performed your measurements. How do your results reflect upon your actual internet link speed and how does your network environment influence your results obtained (provide examples)? Are there outliers in your data?

4.3 Plot the bandwidth-delay product versus the hop count, this time omitting any outliers. Is there any correlation? If not, why not?

4.4 When running your tests for bandwidth, delay, and jitter, were there any variables which may have affected the accuracy or reliability of your results? How might you improve upon these (explain your rationale with examples from your experience)?