High Class

A majority of networks today run TCP/IP, so the odds are good that your network uses IP addresses that you probably inherited from a previous administrator. You might be tempted to simply accept these existing IP addresses. If and when you create new addresses, you might not think to look beyond Classes A, B, or C. However, you can use Class D IP addresses to increase your network's efficiency and benefit multicasting applications.

Unicasting vs. Multicasting and Broadcasting
Every Windows 2000 or Windows NT book you pick up will point out that you can use Class A, B, and C IP-address ranges on your network. But two more classes (i.e., Class D and Class E) offer values in the range beyond 223, which is the upper limit of Class C's first octet.

IP uses unicasting to communicate between addresses in Classes A, B, and C. This method entails a host sending a communication to one other host, as Figure 1, page 102, shows. Usable addresses in Classes D and E use multicasting, which entails a host sending a communication to multiple select hosts, as Figure 2, page 102, shows, or broadcasting, which entails a host sending a communication to all other hosts on the subnet, as Figure 3, page 102, shows. (Class A, B, and C addresses can also employ broadcasting, but not by default.)

Broadcasting and Class E
Broadcasting generates traffic on the Internet and should be used selectively. That said, if you want to send a broadcast to a subnet other than the local subnet, you can use 255 as the ending octet or octets. For example, to send a broadcast to the entire 207.16.99 network, use the address 207.16.99.255. To send a broadcast to the Class B address 133.87, use the address 133.87.255.255.

A Class E address' first octet can range in value from 240 to 255. With rare exceptions, these addresses are reserved for future use and experimentation. The primary exception is 255.255.255.255, which is used to send a LAN broadcast (often to a boot server).

Multicasting and Class D
Multicasting is more efficient than unicasting, and much more efficient than broadcasting, for delivering data to multiple recipients. For example, to send a slide presentation to 300 users on 300 hosts, unicasting sends the slide presentation 300 times. To perform the same task, multicasting sends the slide presentation once to one Class D address that represents a multicast group to which the 300 hosts subscribe. Thus, multicasting reduces the amount of network traffic and the amount of consumed bandwidth.

Any host in any location can subscribe to a multicast group; the subscribing host needn't be on the same subnet as the sending host. A host can subscribe to more than one multicast group. Each host in a group receives the same data as the other hosts in the group. (However, multicast packets might not arrive in the order in which a host sends them, so subscribing hosts might need to compile and sort multiple packets to reassemble the sent data.) Multicasting is the foundation for Multicast Backbone (MBone), videoconferencing, and similar multicast-enabled applications. (For more information about multicasting applications, see Tao Zhou, "Be Prepared for IP Multicasting Applications," May 1997.)

You can use Class D addresses for multicasting. A Class D address' first octet can range in value from 224 to 239. Keep in mind that most addresses in the 224 range are reserved for specific purposes; you shouldn't use these addresses. (Some Class A, B, and C addresses are also reserved. For more information about these addresses, see the sidebar "Other Special Addresses.") Among the reserved Class D addresses are the following:

• 224.0.0.1 (for multicasting to all hosts on a subnet)
• 224.0.0.2 (for multicasting to all routers on a subnet)
• 224.0.0.4 (for multicasting to all Distance Vector Multicast Routing Protocol—DVMRP—routers)
• 224.0.0.5 (for multicasting to all Multicast Open Shortest Path First—MOSPF— routers)
• 224.0.0.9 (for multicasting to all Routing Information Protocol version 2— RIP2—routers)
• 224.0.0.10 (for multicasting to all Interior Gateway Routing Protocol— IGRP—routers)
• Addresses in the range from 224.0.12.0 to 224.0.12.63 (for use by MSNBC)
• Addresses in the range from 224.0.18.0 to 224.0.18.255 (for use by Dow Jones)

You can use multicasting within a private organization to communicate with a selected group of users anywhere. The Internet Engineering Task Force (IETF) and the Internet Assigned Numbers Authority (IANA) recommend that you use an address in the 239.0.0.0 to 239.255.255.255 range for one organization. If you want to multicast over the Internet, you must obtain a multicast address from the IANA.

Multicast Routing
Multicast routing involves sending messages to or among multiple routers for the purpose of updating routing tables. A multicast router uses the Internet Group Management Protocol (IGMP), which defines membership within a multicast group, to examine a multicast packet's header and determine whether to send data to local recipients. You can use Mcast, a diagnostic utility in the Microsoft Windows 2000 Resource Kit, to determine which routers on your network are enabled for IP multicasting. Multicast routers communicate with one another using one or more of the following protocols:

• MOSPF—IETF Request for Comments (RFC) 1584 defines MOSPF. This protocol is one of the best examples of multicast router communication.
• DVMRP—RFC 1075 defines this protocol, which uses Link State Advertisements, so each router can multi-cast available links within the network.
• Protocol Independent Multicast-Sparse Mode (PIM-SM)—RFC 2362 defines this protocol, which replaces RFC 2117. PIM-SM lets unicast routers support multicasting. You typically use PIM-SM in conjunction with IGRP.