7.11 Exercises

1:	Does a gasoline engine have a single point of failure? Does a motorized fire engine? Does a fire department? How does each of the last two compensate for single points of failure in the previous one(s)? Explain your answers.
2:	Telecommunications network providers and users are concerned about the single point of failure in "the last mile," which is the single cable from the network provider's last switching station to the customer's premises. How can a customer protect against that single point of failure? Comment on whether your approach presents a good cost-benefit trade-off.
3:	You are designing a business in which you will host companies' web sites. What issues can you see as single points of failure? List the resources that could be involved. State ways to overcome each resource's being a single point of failure.
4:	The human body exhibits remarkable resilience. State three examples in which the body compensates for failure of single body parts .
5:	How can hardware be designed for fault tolerance? Are these methods applicable to software? Why or why not?
6:	The old human telephone "switches" were quaint but very slow. You would signal the operator and say you wanted to speak to Jill, but the operator, knowing Jill was visiting Sally, would connect you there. Other than slowness or inefficiency, what are two other disadvantages of this scheme?
7:	An (analog) telephone call is "circuit based," meaning that the system chooses a wire path from sender to receiver and that path or circuit is dedicated to the call until it is complete. What are two disadvantages of circuit switching?
8:	The OSI model is inefficient; each layer must take the work of higher layers, add some result, and pass the work to lower layers . This process ends with the equivalent of a gift inside seven nested boxes, each one wrapped and sealed. Surely this wrapping (and unwrapping ) is inefficient. From reading earlier chapters of this book, cite a security advantage of the layered approach.
9:	Obviously, the physical layer has to be at the bottom of the OSI stack, with applications at the top. Justify the order of the other five layers as moving from low to high abstraction.
10:	List the major security issues dealt with at each level of the OSI protocol stack.
11:	What security advantage occurs from a packet's containing the source NIC address and not just the destination NIC address?
12:	TCP is a robust protocol: Sequencing and error correction are ensured, but there is a penalty in overhead (for example, if no resequencing or error correction is needed). UDP does not provide these services but is correspondingly simpler. Cite specific situations in which the lightweight UDP protocol could be acceptable, that is, when error correction or sequencing is not needed.
13:	Assume no FTP protocol exists. You are asked to define a function analogous to the FTP PUT for exchange of files. List three security features or mechanisms you would include in your protocol.
14:	A 32-bit IP addressing scheme affords approximately 4 billion addresses. Compare this number to the world's population. Every additional bit doubles the number of potential addresses. Although 32 bits is becoming too small, 128 bits seems excessive, even allowing for significant growth. But not all bits have to be dedicated to specifying an address. Cite a security use for a few bits in an address.
15:	Sidebar 7-1 on usage of the University of Illinois site states that on a particular day, the site had been visited by hosts in 57 different countries . How do they know? Describe a means that an attacker could make it seem as if there were accesses from 57 countries. In your answer, cover both an actual number smaller than 57 and greater than 57. (You may treat these two cases separately.)
16:	When a new domain is created, for example, yourdomain.com, a table in the .com domain has to receive an entry for yourdomain. What security attack might someone try against the registrar of .com (the administrator of the .com table) during the creation of yourdomain.com?
17:	Describe a social engineering attack you could use to obtain a user 's password.
18:	Is a social engineering attack more likely to succeed in person, over the telephone, or through e-mail? Justify your answer.
19:	A port scanner is a tool useful to an attacker to identify possible vulnerabilities in a potential victim's system. Cite a situation in which someone who is not an attacker could use a port scanner for a nonmalicious purpose.
20:	One argument in the security community is that lack of diversity is itself a vulnerability. For example, the two dominant browsers, Netscape Navigator and Microsoft Internet Explorer, are used by probably 95 percent of Internet users. What security risk does this control of the market introduce? Suppose there were three (each with a significant share of the market). Would three negate that security risk?
21:	Compare copper wire, microwave, optical fiber, infrared, and (radio frequency) wireless in their resistance to passive and active wiretapping.
22:	How many 4-digit numeric PINs are there? Suppose a ban prohibits "obvious" PINs like 0000 and 1234. What would you define as "obvious," and how many PINs would that disallow? Now assume people choose PINs not for their numeric significance but because they are all or the first four letters of a word as on a telephone keypad: A,B,C=2, D,E,F=3, etc. Using the cryptanalysis lessons from Chapter 2, estimate how many 4-digit PINs that people might choose would really exist. The goal of this exercise is not to obtain the definitive answer but to present a reasonable analysis. In what network settings is a PIN still justified as an authenticator, in spite of the small numbers you obtain in this exercise?
23:	What is a "man-in-the-middle" attack? Cite a real-life example (not from computer networking) of such an attack. Suggest a means by which sender and receiver can preclude a man-in-the-middle attack. (a) Cite a means not requiring cryptography. (b) Cite a means involving cryptography but also ensuring that the man in the middle cannot get in the middle of the key exchange.
24:	Suggest a countermeasure for traffic flow analysis.
25:	A problem with pattern matching is synonyms. If the current directory is bin, and . denotes the current directory and .. its parent, then bin, ../bin, ../bin/., .././bin/../bin all denote the same directory. If you are trying to block access to the bin directory in a command script, you need to consider all these variants (and an infinite number more). Cite a means by which a pattern-matching algorithm copes with synonyms.
26:	The HTTP protocol is by definition stateless, meaning that it has no mechanism for "remembering" data from one interaction to the next. (a) Suggest a means by which you can preserve state between two HTTP calls. For example, you may send the user a page of books and prices matching a user's query, and you want to avoid having to look up the price of each book again once the user chooses one to purchase. (b) Suggest a means by which you can preserve some notion of state between two web accesses many days apart. For example, the user may prefer prices quoted in euros instead of dollars, and you want to present prices in the preferred currency next time without asking the user.
27:	How can a web site distinguish between lack of capacity and a denial-of-service attack? For example, web sites often experience a tremendous increase in volume of traffic right after an advertisement with the site's URL is shown on television during the broadcast of a popular sporting event. That spike in usage is the result of normal access that happens to occur at the same time. How can a site determine that high traffic is reasonable?
28:	Syn flood is the result of some incomplete protocol exchange: The client initiates an exchange but does not complete it. Unfortunately, these situations can also occur normally. Describe a benign situation that could cause a protocol exchange to be incomplete.
29:	A distributed denial-of-service attack requires zombies running on numerous machines to perform part of the attack simultaneously . If you were a system administrator looking for zombies on your network, what would you look for?
30:	Signing of mobile code is a suggested approach for addressing the vulnerability of hostile code. Outline what a code-signing scheme would have to do.
31:	The system must control applets' accesses to sensitive system resources, such as the file system, the processor, the network, and internal state variables . But the term "the file system" is very broad, and useful applets usually need some persistent storage. Suggest controls that could be placed on access to the file system. Your answer has to be more specific than "allow all reads" or "disallow all writes ." Your answer should essentially differentiate between what is "security critical" and not, or " harmful " and not.
32:	Suppose you have a high-capacity network connection coming into your home, and you also have a wireless network access point. Also suppose you do not use the full capacity of your network connection. List three reasons you might still want to prevent an outsider from obtaining free network access by intruding into your wireless network.
33:	Why is segmentation recommended for network design? That is, what makes it better to have a separate network segment for web servers, one for the back-end office processing, one for testing new code, and one for system management?
34:	For large applications, some web sites use devices called "load balancers" to distribute traffic evenly among several equivalent servers. For example, a search engine might have a massive database of content and URLs, and several front-end processors that formulate queries to the database manager and format results to display to an inquiring client. A load balancer would assign each incoming client request to the least busy processor. What is a security advantage of using a load balancer?
35:	Can link and end-to-end encryption both be used on the same communication? What would be the advantage of that? Cite a situation in which both forms of encryption might be desirable.
36:	Does a VPN use link encryption or end-to-end? Justify your answer.
37:	Why is a firewall a good place to implement a VPN? Why not implement it at the actual server(s) being accessed?
38:	Does a VPN use symmetric or asymmetric encryption? Explain your answer.
39:	Does a PKI perform encryption? Explain your answer.
40:	Does a PKI use symmetric or asymmetric encryption? Explain your answer.
41:	Should a PKI be supported on a firewall (meaning that the certificates would be stored on the firewall and the firewall would distribute certificates on demand)? Explain your answer.
42:	Why does a PKI need a means to cancel or invalidate certificates? Why is it not sufficient for the PKI to stop distributing a certificate after it becomes invalid?
43:	Some people think the certificate authorities for a PKI should be the government, but others think certificate authorities should be private entities, such as banks, corporations, or schools . What are the advantages and disadvantages of each approach?
44:	If you live in country A and receive a certificate signed by a government certificate authority in country B, what conditions would cause you to trust that signature as authentic ?
45:	A certificate contains an identity, a public key, and signatures attesting that the public key belongs to the identity. Other fields that may be present include the organization (for example, university, company, or government) to which that identity belongs and perhaps suborganizations (college, department, program, branch, office). What security purpose do these other fields serve, if any? Explain your answer.
46:	What is the security purpose for the fields, such as sequence number, of an IPSec packet?
47:	Discuss the trade-offs between a manual challenge response system (one to which the user computes the response by hand or mentally) and a system that uses a special device, like a calculator.
48:	A synchronous password token has to operate at the same pace as the receiver. That is, the token has to advance to the next random number at the same time the receiver advances. Because of clock imprecision, the two units will not always be perfectly together; for example, the token's clock might run 1 second per day slower than the receiver's. Over time, the accumulated difference can be significant. Suggest a means by which the receiver can detect and compensate for clock drift on the part of the token.
49:	The workstation is a weak link in systems like PKI and Kerberos: A compromised workstation can collect and transmit cleartext passwords and encryption keys. Suggest a means to prevent compromise of a workstation from, for example, a Trojan horse.
50:	This chapter listed several disadvantages to ACLs on routers as a network access control method. List two advantages.
51:	List a situation in which you might want to block (reject) certain traffic through an ACL on a router; that is, a situation in which the performance penalty would not be the deciding factor.
52:	What information might a stateful inspection firewall want to examine from multiple packets?
53:	Recall that packet reordering and reassembly occur at the transport level of the TCP/IP protocol suite. A firewall will operate at a lower layer, either the Internet or data layer. How can a stateful inspection firewall determine anything about a traffic stream when the stream may be out of order or damaged?
54:	Do firewall rules have to be symmetric? That is, does a firewall have to block a particular traffic type both inbound (to the protected site) and outbound (from the site)? Why or why not?
55:	The FTP protocol is relatively easy to proxy; the firewall decides, for example, whether an outsider should be able to access a particular directory in the file system and issues a corresponding command to the inside file manager or responds negatively to the outsider. Other protocols are not feasible to proxy. List three protocols that it would be prohibitively difficult or impossible to proxy. Explain your answer.
56:	How would the content of the audit log differ for a screening router versus an application proxy firewall?
57:	Cite a reason why an organization might want two or more firewalls on a single network.
58:	Firewalls are targets for penetrators. Why are there few compromises of firewalls?
59:	Should a network administrator put a firewall in front of a honeypot? Why or why not?
60:	Can a firewall block attacks using server scripts, such as the attack in which the user could change a price on an item offered by an e-commerce site? Why or why not?
61:	Why does a stealth mode IDS need a separate network to communicate alarms and to accept management commands?
62:	One form of IDS starts operation by generating an alert for every action. Over time, the administrator adjusts the setting of the IDS so that common, benign activities do not generate alarms. What are the advantages and disadvantages of this design for an IDS?
63:	Can encrypted e-mail provide verification to a sender that a recipient has read an e-mail message? Why or why not?
64:	Can message confidentiality and message integrity protection be applied to the same message? Why or why not?
65:	What are the advantages and disadvantages of an e-mail program (such as Eudora or Outlook) that automatically applies and removes protection to e-mail messages between sender and receiver?