Chapter 20

Review Questions

1. Which packet parameters can be used when traffic filtering on the router:

• Source IP address - Correct
• Protocol carried in the IP packet - Correct
• UDP/TCP port number - Correct
• Source IP address of the previous packet - Typo in the question, should be 'IP address of the previous node on the route '

2. What is the difference between the routing advertisements and the results of filtering user traffic?

Both kinds of filtering result in some kind of restrictions in connectivity between network nodes as some packets sent won't be delivered to destinations. The aim of the routing  advertisments filtering is to make some networks or subnetworks hidden to ather routers and hence the users of other networks can't communicate to these networks/subnetworks nodes at all. When a router filters user packets the aim is a bit different, it is to selectively protect network nodes from receiveing particular kind of packets, e.g. packets that carry ICMP echo requests which can cause normal service disruption.

3. What is the meaning of the word "integrated" in the name of the IntServ technology?

This word reflects the fact that the IntServ technology describes all the elements of QoS support as a tightly integrated system

4. What parameter makes it possible to limit the burst of the input packet flow profiled using the Token Bucket algorithm?

The parameter b that corresponds to the volume of the token bucket

5. Why does the probability of discarding packets in the random early detection (RED) method depend on the averaged queue length rather than on the current queue length?

To make the behaviuor of RED more stable as in this case short-lived fluctuation of the queue length doesn't change instantly the queue processing

6. Why the RED mechanism is not applicable for the UDP traffic?

Because UDP doesn't support congestion avoidance mechanism which RED exploits

7. Explain the main stages of router resource reservation using RSVP.

8. What is the principal limitation of the IntServ technology:

• It cannot be applied to multicast addressing
• The router must store information on the state of each link - Correct
• End nodes must periodically update the reservation - Correct

9. Why doesn't the DiffServ technology use a signaling protocol?

A signalling protocol normally is used for dynamic resource resrvation for an individual data flow. DiffServ deals with traffic classes and not with indovidual flows; in this case a reservation can be easily made statically whothout dynamic protocol.

10. What is the difference between EF and AF services?

EF service (more precisely - EF-based service as EF defines per-hop behaviuor and can be used for building different services) is design to support the most delay-sensitive traffic, e.g. VoIp traffic. AF service provides traffic with guaranteed delivery (with different levels of probability) and guaranteed bandwidth which makes AF suitable for drop-sensitive traffic which requires certain share of bandwidth, e.g. multimedia streaming. AF also decreases the probability of packet delays during congestion periods but not to such extent as EF does.

11. What specific features of the DiffServ technology made it popular among communications carriers:

• It can be implemented within the limits of the communications carrier network independently of the networks of other communications carriers - Correct
• Routers operate with traffic classes, which does not create a significant load on routers- Correct
• It ensures automation of QoS computations - Incorrect

12. What is the main goal of NAT?

• To thwart DoS attacks - Incorrect
• To solve the problem of IP address shortage in IPv4 - Correct
• To protect the internal address space of the company - Correct

13. Which additional packet attributes are used in NAT for mapping the set of internal addresses to a single global address?

TCP or UDP port numbers

14. Fill in the "Designated port" column of the NAT table.

Any unique for the given global IP address number can be used, e.g.:

15. List the main variants of router architecture.

16. What are criteria used for classifying routers?

17. What are the specific features of Layer 3 switches?

Problems

1. Compose an access list or lists for a Cisco router connecting the company to the Internet (Fig. 20.13). The access list must ensure the following:

• Communications of users of network 194.100.12.0/24, except for the user 194.100.12.25, with only the nodes of networks 132.22.0.0/16 and 201.17.200.0/24. These users must not exchange information with
  the Internet.
• User 194.100.12.25 must be provided with the possibility of unlimited
  data exchange.
• Servers of network 201.17.200.0/24 must be accessible from the Internet only using ftp and http. Access to these servers using ICMP must be denied.

Fig. 20.13. Traffic filtering using a router

access-list 101 permit  IP host 194.100.12.25 any

access-list 101 permit  IP 194.100.12.0 0.0.0.255 132.22.0.0 0.0.255.255

access-list 102 deny ICMP any 201.17.200.0 0.0.0.255 eq 8

!the command above is not mandatory as the following two command permit only two kinds of TCP packets, therefore ICMP packets will be discarded as not permitted explicitely. However, the command in question is useful as some other kinds of packets might be permitted in the future and this might implicitely permit ICMP.

access-list 102 permit TCP any 201.17.200.0 0.0.0.255 21

access-list 102 permit TCP any 201.17.200.0 0.0.0.255 80 

interface s1

access-group 101 out

interface e1

access-group 102 out

2. When describing the NAT technology, we simplified the pattern. In particular, we didn't consider the problems that might arise when ICMP error messages arrive in the internal network. Suggest your own variant of the algorithm that should be used by the NAT protocol when an ICMP message arrives at its external interface. Tip: Before passing the ICMP message, ICMP must introduce corrections not only into the IP header but also into the ICMP data field.