1



Department of Computer Science and Engineering

CSE 3214: Computer Network Protocols and Applications

Final Examination

Instructor: N. Vlajic



Instructions:

• Examination time: 180 min.
• Print your name and CS student number in the space provided below.
• This examination is closed book and closed notes. Use of calculators is allowed.
• There are 9 questions. The points for each question are given in square brackets, next to the
question title. The overall maximum score is 100.
• Answer each question in the space provided. If you need to continue an answer onto the back
of a page, clearly indicate that and label the continuation with the question number.








FIRST NAME: ___________________________

LAST NAME: ___________________________

STUDENT #: ___________________________



Question Points
1 / 10
2 / 10
3 / 12
4 / 10
5 / 16
6 / 12
7 / 9
8 / 14
9 / 7
Total / 100



2
1. Multiple Choice, True/False [10 points]

Multiple choice questions – each question is worth [1 point].

a) What is the limited broadcast address corresponding to the node with the following IP
address: 131.15.46.59?
(a) 131.15.46.255
(b) 131.15.255.255
(c) 255.255.255.255
(d) None of the above.

b) In hexadecimal colon notation, a 128-bit long IPv6 address is divided into ____________
sections, each comprising ____________ hexadecimal digits
(a) 4 : 2
(b) 8 : 4
(c) 16 : 2
(d) none of the above

c) An ARP reply is normally ____________ .
(a) broadcast
(b) multicast
(c) unicast
(d) none of the above

d) When the hop-count field in an IP packet reaches zero and the destination has not been
reached, an ICMP ____________ error message is sent back to the sending machine.
(a) destination-unreachable
(b) time-exceeded
(c) parameter-problem
(d) none of the above

e) Which statement(s) is true regarding the advantages of static over dynamic routing?
(a) reduces the probability of packet drop due to router (i.e. route) failures
(b) easier to implement and maintain in large, complex networks
(c) increased security
(d) none of the above




3
f) A port address in UDP is ____________ bits long.
(a) 8
(b) 16
(c) 32
(d) none of the above

g) The ports ranging from 1,024 to 49,151 are called ____________ .
(a) well-known
(b) dynamic
(c) registered
(d) none of the above

h) In TCP, Clark’s solution can solve the silly window syndrome created by the ____________ .
(a) sender
(b) receiver
(c) both sender and receiver
(d) none of the above

i) The ____________ utility allows you to query the DNS database from any computer on the
network and find the host name of a device by specifying its IP address, or vice versa?
(a) ipconfig
(b) tracert
(c) nslookup
(d) netstat

j) A user needs to send the Web-server some information. The respective HTTP request line
method is ____________ .
(a) GET
(b) HEAD
(c) SEND
(d) POST

4
2. Packet Switching [10 points]

Consider two serially connected packet switches as shown in the figure below. The link connecting
the two switches (Link 1) is full-duplex, with datarate of 1 Mbps and propagation delay of 3 ms (in
each direction). The packets sent through the network are 1000 bits long. The input buffer/queue of
Switch 2 can store at most 100 packets.

To control congestion and avoid packet loss, the switches employ the so-called ‘back pressure’
mechanism. According to this mechanism, whenever Switch 2 detects congestion on its outgoing link
(i.e., no more packets can be sent over Link 2), Switch 2 sends a signal back to Switch 1 instructing
Switch 1 to halt further packet transmission over Link 1.

In this question, you are asked to determine how big Switch 2 should let its buffer grow (in case of
congestion on Link 2), before sending a back pressure signal to Switch 1. Your answer should be in
the units of ‘packets’.



Additional assumptions:
• There is an unlimited number of packets at Switch 1; therefore, when active, Switch 1 sends
packet continuously, back-to-back.
• Once detected, the congestion on Link 2 could be alleviated at any point in time. Hence, the back
pressure signal should not be sent too early (unless there is a real risk of Switch 2 running out of
buffer space), nor too late (no packet should ever be lost/dropped).

Link 1 Link 2

5
3. NAT and IPv6 [12 points]

3.1) [4 points] Assume an IP packet carrying an HTTP request is going from a local (i.e.
home) area network onto the wider Internet through a NAT router. Name all header fields that the
NAT router needs to change in the given packet? Explain your answer.

(Hint: encapsulation as well as the syntax/semantics of all involved protocols must be taken into
consideration.)













3.2) [4 points] The diagram below shows a packet traveling through a NAT router. Packet 1 is
sent from the internal host (S) to the NAT router, packet 2 is sent from the NAT router to the external
web server (D), packet 3 is received from the web server by the NAT router, and packet 4 is sent by
the NAT router to the original host.

Fill in the missing source and destination IP addresses and port numbers in packets 2-4.




6
3.3) [4 points] The diagram below shows an IPv6 packet tunnelled over IPv4. Fill in the
missing source and destination addresses at places/packets marked 1, 2, and 3.









3.4) [4 points] Name two modifications of IPv6, from the IPv4, that allow a router to process a
packet quicker.
1 2 3

7
4. Routing [10 points]

4.1) [3 points] What are the differences between routing and forwarding? Briefly explain each
of them.












4.3) [7 points] Consider the network configuration shown in figure below. Assume that each
link has the same cost.



a) Run Bellman-Ford algorithm on this network to compute the routing table for node A. Show A's
distance to all other nodes at each step.






8

b) Suppose the link A-B goes down. As a result, A advertises a distance of infinity to B. Describe in
detail a scenario where C takes a long time to learn that B is unreachable


9
5. TCP Potpourri [16 points]

5.1) [5 points] Suppose you want to send a file of size 255,000 over a 2 Mbps link using TCP.
The maximum segment size (MSS), which represents the size of TCP payload, is 1,000 bytes. Two-
way propagation delay between the source and the destination is 10 [msec] . TCP Threshold = 130
packets.

How long will it take to transmit the given file, from the start to the end of TCP transfer?





























5.2) [6 points] Consider a TCP connection that, at time 0, is in the congestion avoidance
phase with a window size equal to 6 MSS, and then sends packets {1, 2, 3, .., 6}. (The connection
has sent packet before that were properly acknowledged. The packet transmission time of a single
packet is Ptransmission=1 sec, while RTT = 8 sec and timeout = 24 sec.) The sender does not implement
fast retransmit and fast recovery, and after every timeout it restarts in slow start.

Now, assume packets 1 and 2 are lost. Draw a timing diagram that shows the transmission of the first
8 packets by completing the figure below. Indicate the significant events, such as start of slow start
(SS), or of congestion avoidance (CA), and the relevant congestion window sizes.

Also, calculate how long, overall, it takes to transmit those 8 packets.

10

















5.3) [5 points] Repeat the previous problem assuming that the host now uses fast retransmit
and fast recovery.




11
6. Socket Programming [12 points]

6.1 [2 points]
Consider the following Java application:

socket = new DatagramSocket(12345);
while (true) {socket.receive(packet);}

What happens if somebody decides to run two instances of this application on one machine and 4
UDP segments arrive at port 12345?
a) both instances of the application receive all 4 segments
b) one instance receives all 4 segments
c) some segments are received by one instance, other segments are received by the other instance
d) one instance receives segments 1 and 3, the other receives segments 2 and 4


6.2 [2 points]
Given the following lines from a Java program segment:

byte[] dataOut = new byte [512];
String userInput = inFromUser.readLine();
dataOut = userInput.getBytes();

which of the below lines of code should be used to create a new UDP datagram packet and send the
data that was provided by the user to a host identified by the InetAddress object IPAddress?
a) DatagramPacket packetToSend = new DatagramPacket(dataOut, dataOut.length, IPAddress)
b) UDPPacket = new UDPDatagram (userInput, userInput.length, IPaddress, 9876)
c) DatagramPacket packet = new DatagramPacket(dataOut, dataOut.length, IPAddress, 9876)
d) Socket datagramSocket = new dataGramPacket(dataOut, IPAddress)


6.3 [2 points]
Consider a server socket object

socket = ServerSocket(12345);

What does the invocation of socket.accept() return?
a) 'true' if there is a new TCP segment in the socket's buffer, false otherwise
b) a new TCP segment from the socket's buffer (blocks if no segments are available)
c) 'true' if a new TCP connection request has arrived, false otherwise
d) a socket associated with a new TCP connection (blocks if no connections are available)


6.4 [2 points]
Assume the following two lines of code are to be executed on the machine blue.cse.yorku.ca:

Socket myFirstSocket = new Socket("blue.cse.yorku.ca", 5555);
Socket mySecondSocket = new Socket(5555);


12
Are the two lines/commands in conflict? Explain briefly!














6.5 [4 points]
Suppose application A is using a UDP socket, i.e. DatagramSocket(), to transfer data to application B
on a remote host. Suppose application A calls send() method on the given socket 10 times.
a) Can the underlying network stack transmit more than 10 data packets?
b) Can the underlying network stack transmit fewer than 10 data packets?
Explain briefly!



13
7. DNS [9 points]

Consider the DNS topology in the figure below, in which different DNS zones are indicated with a
dashed line. There is only one DNS server per each zone, and it happens to have the same name as
the highest node in that zone: yale.edu, cs.yale.edu, ruthgers.edu, cs.rutgers.edu, and root-
servers.net. The only DNS servers supporting recursive querying are cs.yale.edu and cs.rutgers.edu.




For each of the queries below, list in order all the DNS servers contacted by the resolver (located in
the OS of the machine running the query). Assume there is no caching performed at any level of the
hierarchy.

6.1 [3 points] A machine called lab1.bio.yale.edu exists in the biology department at Yale,
and a user on eden.rutgers.edu launches this query: "nslookup lab1.bio.yale.edu".










6.2 [3 points] At the prompt of paul.cs.rutgers.edu somebody launches this query:
"nslookup lab1.bio.yale.edu".










14

6.3 [3 points] On lab1.bio.yale.edu somebody queries: "nslookup paul.cs.rutgers.edu".


15
8. HTTP [14 points]

8.1) [6 points] Suppose that a browser on a host A wants to retrieve an HTML document (D),
and an embedded image (I), from a host B. Assume that A does not initially now the IP address of B,
but A's local name server S does know B's IP address. Also, assume that the browser on A uses
HTTP/1.0 (the non-persistent version).

Show the chronological sequence of transport layer segments (TCP or UDP) sent and the respective
application layer data type (DNS or HTTP) by filling in the following table. Also, for each TCP packet
state when any of the SYN, FIN and/or ACK bits in the TCP header are set to 1.

Source
Machine
Destination
Machine
Transport Layer
Protocol
Application Layer
Protocol























16
8.2) [8 points] An institution has a T1 access line (1.544 Mbps) which services web requests
from about 150 users in the network. The institution is considering putting up a proxy server which
can act as a cache for all these users. (I.e., whenever a user makes a web request, the request is first
sent to the proxy server to see if it can service the request from its cache. Otherwise, the proxy
forwards the requests to the origin server.)

a) [2 points] On average, each user makes 1 request per second, with an average size of a
request object being 10 kbits. What percentage of the bandwidth is consumed by the web traffic when
there is no proxy server, and all requests are serviced by the origin server?










b) [3 points] After installing the proxy server, it was found that the hit rate to the cache is
40%. What percentage of the institutional bandwidth is now consumed by the Web traffic?










c) [3 points] If the round trip time (RTT) from a user node to the proxy server is 50 msec,
and the RTT from the proxy server into the Internet is 2 sec, then what is the average delay
experienced by web requests in scenario b)?



17
9. Security [7 points]

9.1) [4 points] Consider one security system using a Key Distribution Center (KDC) server,
and another security system using Certification Authority (CA) server.
a) Suppose the KDC goes down. What is the impact on the ability of parties in the respective system
to communicate?
b) Now suppose that the CA goes down. What is the impact of this failure instead?

Briefly justify your answer, by clearly stating which of the two events would have more serious
consequences.

















9.2) [3 points] We have discussed the concept of Reflector DDoS in class. Do you think it
would be possible to conduct such an attack using the DNS protocol? Briefly justify your answer.

18
IPv4 and IPv6 PACKET FORMATs:







TCP PACKET FORMAT:



19