CS 553
Spring 2003
Diagnostic Exam

Completing this exam to the best of your ability is required for entry into the course.


This is a take home diagnostic exam. It is due at 1:00PM, Friday, 1/24/2003. Send your answers via email to the instructor (rmartin@cs.rutgers.edu). Use text or PDF formats only. No question should require more than a ½ page long answer.


Work alone. You can use the Internet, or any other non-human resource. If you use an outside resource to answer a question, note it your answer (a URL is fine). Note: Failure to cite outside resources will reflect badly on your credibility.


The purpose of this exam is to let me know how much background you have on the material. Don't let the quality of your answers be inconsistent with your performance during the rest of the class!


If you can't answer most of the questions in 48 hours by yourself, you're not ready to take this class. Please take CS552,graduate computer networks, and CS519,operating systems, first.



Questions:

  1. Explain how the program "traceroute" works. Explain why sometimes it gives incomplete information.

  2. What is a stochastic process? Under what circumstances and conditions are they useful for modeling real systems?

  3. When are stochastic models not an appropriate modeling method?

  4. Explain, using mathematics, how to find the steady-state distribution of a simple discrete time markov chain.

  5. Write the pseudo-code for a client and server connecting to each other over a TCP socket.

  6. From an Internet Service perspective, what is a cookie? Sketch a situation where it makes sense to use cookies.

  7. What is a two-phase commit protocol? Briefly explain what happens in each phase.

  8. What is an articulation point router? Why are these routers of interest over other kinds of routers?

  9. What is a heavy-tailed distribution? Why would we concern ourselves with trying to characterize a distribution of some metric (e.g. file size, request arrival interval) as heavy-tailed or not?

  10. What is an inverted index? How fast does an inverted index grow as the web grows?