No Title

CS 416/516 Computer Communications and NetworksFall 1993Final Exam (December 17)Prof. Thomas Narten

NAME: SCORE:

This is a closed book (and notes) examination; however, you may consult your summary. Answer all questions on the exam itself. Justify your answers! You must explain your reasoning to receive full credit on short answer questions.

All questions have equal weight. However, some questions are harder than others; budget your time accordingly.

Students enrolled in CSI 516 may omit any one question. Students enrolled in CSI 416 may omit any three questions. Students must clearly mark those questions that are not to be graded, or I will assume you wish to omit the final question(s) on the exam.

Consider a small internet of machines connected to Ethernets. Is there a one-to-one mapping between machine names and IP addresses? What about between IP addresses and Ethernet addresses? Explain.
As the network manager for Kludge System, Inc., you are in charge of connecting your company to the Internet. There are currently only a dozen machines in your company, but the number is expected to grow to at least 1000 within five years. You will likely also have multiple networks as the company expects to open new offices in other states. How many and what type (e.g., class A, B or C) of IP addresses would you recommend? Explain.
What purpose does the 16-bit identifier field serve in the IP header? When and how is it used?
Consider a sliding window protocol whose send window is 16 packets. When a single packet loss occurs, the send window cannot advance until after an ACK is received for the retransmission of the lost packet. When the ACK finally arrives, however, it advances the send window by 16 packets (assuming no other packets were lost), causing the sender to generate 16 back-to-back packets. This burst of packets may well lead to congestion and further loss. How does TCP avoid this problem? When (if ever) does TCP send back-to-back packets?
Explain how the IP time-to-live field can be used to determine that current path between a source and destination.
Explain how hosts determine which gateway to use when forwarding packets to a non-local destination. In particular, how is it that hosts need NOT run the full-blown routing protocols gateways do?
What is are autonomous regions and what purpose do they serve?
Suppose someone attempts made to send mail to someone at a remote site (e.g., ``cs.purdue.edu''). Would it be possible for the mail message to be delivered immediately if none of the root DNS servers were currently reachable, or would the message be temporarily delayed? Explain.
Outline the steps a host takes in determining whether a packet it is about to send can be sent directly or needs to go through a gateway. Assume subnetting is in use. Be specific!
Are ports (as used by TCP and UDP) more like mailboxes or process identifiers? Explain.
Describe TCP's three-way handshake. What packets are exchanged, and what are the contents of the relevant packet fields? Be specific!
What purpose does the variable ssthresh (slow start threshold) serve in TCP's congestion control mechanism? How and when is the variable used?
Give two reasons why the IP header contains a time-to-live field.
In class, we discussed how call-by-reference doesn't work well when making RPCs. How is it possible then that SUN RPC allows you to return a pointer to a data structure (e.g., as in project 5)?
What (if any) are the main differences between protecting against replay attacks and surpressing duplicate packets?
What purpose does an authentication server play and when is it used?
One can send electronic mail to ``user@foo.albany.edu'' for any machine ``foo'' in the Albany CS department. However, mail sent to ``user@karp.albany.edu'' does not actually ever visit the machine karp. How can this be?

About this document ...

The command line arguments were:
latex2html -split 0 final.tex.

The translation was initiated by Thomas Narten on Sat Oct 12 13:55:12 EDT 1996

Thomas Narten
Sat Oct 12 13:55:12 EDT 1996