UM EECS 489 Winter 1998 Homework Assignment 2

EECS 489 Fall 1999

Homework Assignment 2

DUE DATE 10/19/1999, 1:30 pm

1.a. (8 points) Each grid represents a memory map of a computer. The column number is added to the row number to obtain the byte address of that location. Label one memory map "big-endian" and the other "little-endian" so that

if a long were stored in bytes 0 through 4 it could be read from left-to-right (i.e., the most significant byte would be on the left.)

Label :

	0	1	2	3
0
4
8
16

	3	2	1	0
0
4
8
16

1.b. (5 points) If the block of memory was initialized as shown below, fill in the memory map of the BIG ENDIAN machine. Use hex notation, 0xXX, to fill in the memory map.

char buf[16] = {0x57, 0x51, 0x99, 0x97,
0x91, 0xab, 0x50, 0x05,
0x41, 0x43, 0x42, 0x44,
0x45, 0x47, 0x46, 0x00};

1.c. (7 points) Suppose the block of memory on both machines were cast to the s_arbitrary data structure given below. Fill the memory map of the LITTLE ENDIAN machine so the members of the s_arbitrary structures on both machines contain the same values. Use hex notation, OxXX, to fill in the memory map.

struct _arbitrary 
{
   long  sequence;
   short id;
   char  version;
   char  hops;
   char[8] str;
} s_arbitrary;

2. This question involves logging in to the EECS or ITD machines. You will likely encounter difficulty in answering this question by solely relying on the CAEN network. As a UM student, you should already have, by default, an ITD account (remotely log into login.itd.umich.edu). If not, contact one of theGSIs immediately. You will still be responsible for answering this question.

2.a. (5 points) Print out and attach a copy of the man page for dig(1) (the 1 indicates that the command is in Section 1 of the man manual).

2.b. Tcpdump is a utility which monitors traffic (e.g., network packets) over a network interface (it is somewhat of a misnomer, since it can also be used to display packets other than TCP). It also, upon request, displays the raw contents of such a packet in hexadecimal notation. The following is an example of part of an IP packet displayed by tcpdump, taken from an actual tcpdump() output:

                         4500 0195 01ef 4000 4006 36c0 8dd5 06f5
                         96d6 d513 041c 0050 41a3 4a34 6a65 7716
                         5018 7d77 afd2 0000 4745 5420 2f61 6761
                         7765 622f 2048 5454 502f 312e 300d 0a52
                         6566 6572 ...

As an example of how to read this packet, the first 16 bits of this IP packet are: 0100 0101 0000 0000 ...

2.b.1. (5 points) What version of IP is this packet?

2.b. 2. (5 points) What is the total length of this packet, in bytes (including header and data)?

2.b.3. (5 points) What is the IP address (in dotted decimal notation) of the machine from which this packet originated?

2.b.4. (5 points) What is the IP address (in dotted decimal notation) of the machine to which this packet is destined?

2.c. (10 points) Using your answers from 2.b.4 and 2.a, find the fully qualified domain name (FQDN) of the machine to which this packet is destined.

The "destination machine" for both parts of question 3 refer to the same machine as the one presented in 2.b.4. and 2c.

3.a. (10 points) Using the /usr/sbin/traceroute (ITD) or /usr/local/bin/traceroute (EECS/DCO) utility (again on the ITD or EECS machines), find the route from your (ITD/EECS) machine to the destination machine in question. Attach a hardcopy of your output.

3.b. (5 points) Can you tell, from the names of the intermediate routers, which cities your packet is going through? If so, which cities is it going through?

4.a. (5 points) Use dig to discover the name servers for the engin.umich.edu domain. (There are multiple authoritative servers.)

4.b. (5 points) How long will the IP addresses of those name servers be cached on the local machine?

4.c. (10 points) Dig @mit.edu for some obscure hostname. Did you get an authoritative answer? How long did it take to get the response? Dig @mit.edu again for the same obscure url. Did you get an authoritative answer a second time? How long did it take to get the response the second time? Explain what you observed.

5. (10 points) One disadvantage of many networks (including the Internet) is that a large percentage of data traffic is sent from source to destination unencrypted. Suppose you are hired by ABC Security, Inc. to add a layer of data encyryption to the OSI 7-layer standard (5-layer as discussed in class). Where would you add this extra layer of security? Explain.

What is one major disadvantage in adding such an extra layer? In other words, why isn't all (Internet) network traffic encrypted? (hint: think of at least one type of data traffic which need not be encrypted, then think of the effects of encryption on this type of traffic).

6. (10 points) Suppose that 10^6 (1 000 000) bits of user data are to be transmitted over a 11-hop path (i.e., there are 11 links between sender and receiver) in a packet-switched network as a series of packets, each containing p data bits and 160 header bits. The bit rate of the lines of b bits per second, and the propagation
delay is negligible. What value of p minimizes this delay?

7. Should the domain name system permit partial match queries (i.e., a wildcard as part of a name) ? Why or why not ? [[Comer 22.14] (3 points)

8. Figure 1 shows a zoning partition of the DNS domain hierarchy.

**Figure 1:** A zoning of the domain hierarchy.
$\begin{figure}\centerline{\epsfig {file=dns.eps}}\end{figure}$

(a): Assuming all name server caches started out empty, list, in order, the name servers contacted when; alps.eecs.umich.edu tries to resolve img.sdl.melco.co.jp. (4 points)
(b): After the above resolution, list, in order, the name servers contacted when
(c): After the above two resolutions, list, in order, the name servers contacted when
(d): After the above three resolutions, list, in order, the name servers contacted when

9. How many IP addresses would be needed to assign a unique network number to every home in [the US]? Is the [IPv4] address space sufficient? (Hint: The US has a population of about 250 millions. State your assumption on the estimate of the number of persons per home on the average.) [Comer 4.7] (8 points)

10.

ifconfig

11. Figure 2 shows the topology of a network with the cost/metric of each link labeled. Draw the minimum spanning tree from node A computed using Dijkstra's shortest path first algorithm. Show the evolution of your P and T, where P is the set of nodes for which shortest path has been computed, and T the list of nodes you are computing shortest path for. (15 points)

**Figure 2:** A network.
$\begin{figure}\centerline{\epsfig {file=spf.eps}}\end{figure}$

12. Consider the network depicted in Figure 2 again. Table 1 shows the routing table entries from each of the nodes for destination D. Each entry consists of the <next_hop, metric> information. Assuming all nodes implement the path hold down algorithm with 2 step hold down, if the links <D, E> and <C,D> break simultaneously,

(a): How many steps does it take for D to be reachable again from all nodes? (10 points)
(b): Show the routing table entries for destination D of all the other nodes when the routing tables stop changing. (5 points)

**Table 1:** Initial Routing Table Entries for Destination D
A	B	C	E	F	G
<D, 1>	<C, 2>	<D, 1>	<D, 1>	<E, 2>	<F, 3>

13. When Class B addresses are exhausted, one could give each AS as many Class C addresses as the number of hosts it wishes to support.

(a): Why does CIDR require that the set of Class C addresses given to an AS be contiguous and of size a power of 2? (5 points)
(b): A more general problem than Class B exhaustion is the total depletion of IP addresses. Do you think CIDR help alleviate this problem? Why or why not? (5 points)

What to Turn in

Turn in your answers to the above questions in class on 10/12/98.