Read appendix 4A of your reading notes (pages 4-109 - 4-115), Integrating Ethernet with a Forwarding Network. You may want to reread Section 4.C-E of the text before reading this assignment, as you will need some familiarity with the details of these sections to understand this appendix.

This reading begins with an overview of a more modern version of Ethernet that what was described in the Metcalfe and Boggs paper we read last week, focusing on how broadcast packets are handled. It then summarizes the issues involved in coupling a broadcast local area link layer with a wide area, packet-switched network layer, introducing the address resolution protocol (ARP) to help do this. ARP makes it possible to discover the link layer identifier corresponding to a network layer address (for example, the Ethernet address of a PC or router corresponding to a given IP address.)

Think about the following questions as you read:

• Compare the packet format of today's Ethernet to the packet format described in the paper from last week. What is different? Why?

• ARP relies on the ability of the link layer to broadcast messages. Suppose the link layer had no broadcast capability. How would you implement ARP?

• The last paragraph of the reading mentions "reverse ARP", which can provide a network address given a link layer address. Why might this be useful?

• Notice that ARP provides no means of verifying that the machine that replies to an ARP request can or will actually route to the requested network address. Think of some ways in which a malicious user might use this to interrupt service or spy on network traffic. Surprisingly, these kinds of network attacks are not considered a particularly severe threat to network security; why do you suppose this is the case?

• Is ARP a reliable protocol? Why or why not?