Chapter 17

17.1	We've covered the following packet formats, each of which has a checksum in its corresponding header: IP, ICMP, IGMP, UDP, and TCP. For each one, describe what portion of an IP datagram the checksum covers and whether the checksum is mandatory or optional.
A:	All are mandatory except the UDP checksum. The IP checksum covers only the IP header, while the others start immediately after the IP header.
17.2	Why do all the Internet protocols that we've discussed (IP, ICMP, IGMP, UDP, TCP) quietly discard a packet that arrives with a checksum error?
A:	The source IP address, source port number, or protocol field might have been corrupted.
17.3	TCP provides a byte-stream service where record boundaries are not maintained between the sender and receiver. How can applications provide their own record markers?
A:	Many Internet applications use a carriage return and linefeed to mark the end of each application record. This is NVT ASCII coding (Section 26.4). An alternative technique is to prefix each record with a byte count, which is used by the DNS (Exercise 14.4) and Sun RPC (Section 29.2).
17.4	Why are the source and destination port numbers at the beginning of the TCP header?
A:	As we saw in Section 6.5, an ICMP error must return at least the first 8 bytes beyond the IP header of the IP datagram that caused the error. When TCP receives an ICMP error it needs to examine the two port numbers to determine which connection the error corresponds to, so the port numbers must be in the first 8 bytes of the TCP header.
17.5	Why does the TCP header have a header length field while the UDP header (Figure 11.2) does not?
A:	There are options at the end of the TCP header, but there are no options in the UDP header.