TCP IP Illustrated

The time of day is a bad choice especially if all the

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: he broadcast address is also output. file:///D|/Documents%20and%20Settings/bigini/Docu...homenet2run/tcpip/tcp-ip-illustrated/append_d.htm (8 of 20) [12/09/2001 14.48.04] Appendix D: Solutions to Selected Exercises 12.4 Berkeley-derived implementations do not allow a broadcast datagram to be fragmented. When we specified the length of 1472 bytes, the resulting IP datagram was exactly 1500 bytes, the Ethernet MTU. Refusing to allow a broadcast datagram to be fragmented is a policy decision-there is no technical reason (other than a desire to reduce the number of broadcast packets). 12.5 Depending on the multicasting support in the various Ethernet interface cards in the 100 hosts, the multicast datagram can be ignored by the interface card, or discarded by the device driver. Chapter 13 13.1 Use some host-unique value when generating the random value. The IP address and link-layer address are two values that should differ on every host. The time-of-day is a bad choice, especially if all the hosts run a protocol such as NTP to synchronize their clocks. 13.2 They added an application protocol header that included a sequence number and a timestamp. Chapter 14 14.1 A resolver is always a client, but a name server is both a client and server. 14.2 The question is returned, which accounts for the first 44 bytes. The single answer occupies the remaining 31 bytes: a 2-byte pointer for the domain name (i.e., a pointer to the domain name in the question), 10 bytes for the fixed-size fields (type, class, TTL, and resource length), and 19 bytes for the resource data (a domain name). Notice that the domain name in the resource data ( doesn't share a suffix with the domain name in the question ( so a pointer can't be used. 14.3 Reversing the order means using the DNS first, and if that fails, trying to convert the argument as a dotted-decimal number. This means every time a dotted-decimal number is specified, the DNS is used, involving a name server. This is a waste of resources. 14.4 Section 4.2.2 of RFC 1035 specifies that a 2-byte length pr...
View Full Document

This test prep was uploaded on 04/04/2014 for the course ECE EL5373 taught by Professor Guoyang during the Spring '12 term at NYU Poly.

Ask a homework question - tutors are online