[cs615asa] HW4 revisited

[Jan Schaumann]

Hello,

I'll be sending out grades for HW4 later today or tomorrow, but here are
a few comments relating to the assignment and your submissions.


The main task of the assignment was to show that you understood what
traceroute does, how it works, and that you thought about the results.
Independent thinking and analysis rather than rote execution of commands
and copy and paste of the output where the objective.  Stating that
something interesting happened, but you have no idea why is not quite
what I had in mind when I asked you to explain your findings.

As the assignment required you to run tcpdump(8), I expected you to (a)
actually do that and (b) show the relevant parts.  That is, you should
be able to identify the _relevant_ packets in the tcpdump(8) output.
The packets of your ssh connection to EC2, for eample, are not what I
was looking for.

You should have observed that a few of the targets could not (reliably)
be reached, or that a few of the hops in between the source and
destination might be shown as '*' in the traceroute(8) output.  I would
have liked you to explain why that might be and what, if anything could
be done to work around this.

Since the traceroute(8) manual page discusses this, using a different
protocol to send the traceroute probes, for example, would have been a
good idea.

The targets listed in the assignment were not random strings I
generated, but specifically chosen.  You should not replace them with
names you liked better or thought more convenient.  You should also have
traced all of them.

In the category of "noteworthy or interesting" things you should have
observed and then explained might fall:

- www.stevens.edu traces to Seattle; Stevens in located in Hoboken, NJ.
  What does this tell you?  (Have you tried tracing www.cs.stevens.edu?)

- the trace to www.du.ac.in appears to bounce across the Atlantic before
  ending in Canada - what gives?

- some of the measurements in between individual hops are in the sub-ten
  milliseconds, while some are in the hundreds - why is that?

- several of you reported that some of the destinations appear to crash
  your browser in the visual traceroute - why is that?

Finally, and this may be the most interesting question:

- many of the traces that leave the US and have to go across either the
  Atlantic or the Pacific ocean appear to jump from within the country
  to the next continent, for example from Colorado; this appears to be
  contrary to what we discussed in the context of submarine cables --
  can you explain this?

I will allow this last question as an extra-credit assignment worth 10
points:

Review/research the concept of 'peering', and identify who the various
network block owners are where the connections appear to 'jump'.  Then
try to explain where/how the packets are travelling.

Write up your findings in clear, spell- and grammar-checked English and
send them to me in a well-formatted plain ascii text file (NOT a .doc,
NOT a .docx, NOT a .pdf, NOT a .jpg, NOT a .mov, NOT a .xls, NOT a
.html, ...) named after your Stevens username and created on
linux-lab.cs.stevens.edu using a standard unix text editor in an email
with the subject "[CS615] traceroute" before 2015-04-13 16:00 EDT.

-Jan