CS 422 (Park) Assignment IV Due: Mar. 28 (Wed.), 2007
Submission instructions: Please type your answers and submit electronic copies using turnin by 5pm on the due date.
You may use any number of word processing software (e.g., Framemaker, Word, L
A
T
E
X), but the final output must be
in pdf or ps format that uses standard fonts (a practical test is to check if the pdf/ps file prints on a CS Department
printer without problem). For experiments and programming assignments that involve output to terminal, please use
script to record the output and submit the output file. Use gnuplot to plot graphs. Use ps2gif to convert a eps/ps
plot to gif format (e.g., for inclusion in Word) if there is a need.
PROBLEM 1
Read Chapters 8, 9, and 10 from Comer.
PROBLEM 2 (40 pts)
The purpose of this problem is to introduce “sniffing” raw data from the “wire” (more precisely Ethernet interface
of sender/receiver PCs) on the machines in the lab. Execute the following command on sender and receiver PCs to
capture Ethernet frames:
%sudo /usr/local/etc/tcpdumpwrap-eth1 -c18 -wEX
The command will capture 18 packets from the Ethernet LAN and save them in the file /var/tmp/login-EX, where
login is your login ID. Generate your own traffic on the private network by doing ping from sender machine to receiver
machine (ping -c count receiver-IP-addr) where count is the number of packets transmitted. What is the value
of count that you need to set so that tcpdumpwrap captures 18 packets? Submit the log file in hexadecimal format.
Also, save your ping terminal interaction using script and submit it along with the hexadecimal tcpdumpwrap log
file.
Using the Ethernet header format(s) discussed in class, decode from the hexadecimal dump what the values for the
fields in the Ethernet header are. Compare these values across the 18 captured frames. Use /sbin/ifconfig to
compare the Ethernet addresses that you have captured with those printed out by ifconfig. Compare the frames
captured at the transmitter with those captured at the receiver. From the value of the type/length field determine
whether the Ethernet frames are DIX- or IEEE 802.3-compliant. Explain how you are able to distinguish between the
two. What is the default payload size of ping? How long is the payload of the Ethernet frame? Are they consistent?
(The default IP header size is 20 bytes.) Noting that the first four bits of an IP header indicate its version number
(4 or 6), locate the version number bits in the Ethernet payload and identify the IP version running on your test
machine. Noting that the last 8 bytes of a 20-byte IP header represent the 4-byte IP source address and 4-byte
IP destination address, respectively, locate the IP source/destination address bits in the payload and compare their
values to those output by ifconfig. What is the content of the payload generated by ping?
PROBLEM 3 (40 pts)
Reimplement the client/server network application in Problem 3, Assignment III, such that UDP (SOCK DGRAM
option in socket()) is used in place of TCP. Unlike TCP, UDP does not retransmit lost packets, its main function
being adding port numbers and IP addresses to payload. Although UDP can be used with write() and read()
systems calls, please use the UDP-specific system calls sendto() and recvfrom() in your client/server code. Keep
in mind that, although unlikely (due to on average low traffic), UDP packets may go missing in the lab’s Ethernet
LAN. Perform the same benchmarks as in Problem 3, Assignment III, and submit the resultant output.
PROBLEM 4 (20 pts)
As a continuation of Problem 2, run tcpdumpwrap-eth1 at the receiver-side (i.e., client-side) of the client/server
network application (before running the TCP-based client/server application) so that the request and response pack-
ets are captured. In the hexadecimal dump, identify the request and response packets by their IP and Ethernet
