Internet Address or IP Address

2009 Mar 07

The Internet Corporation For Assigned Names and Numbers (ICANN) sets policy and assigns value for names and otehr constants used in protocols as well as addresses. Usually origanizations and inidividuals interacst with their Internet Service provider (ISP) to get IP addresses, who in turn interact with larger ISPs who then interact with ICANN. See RFCs 1700 and 1117.

IPV4 Network Id

Each host on a TCP/IP internet is assigned a unique 32 bit internet address that is used in all communications with or by that host. This address is chosen in a way to make routing easier such that all hosts on the same network have the same prefix. Conceptually each address is a pair: netid, histid.

In actuality an internet address canot uniquely identifie a hosts network connection because a router (for example) is connected to multipel networks. Such systems are called multi-homed hosts and require multiple IP addresses, one per network connection. Thus an IP address identifies a connection to a network, not an individual system.

By convention hostid 0 is never assigned to a host connection. An IP address with hostid 0 is used to refer to the network itself, or it used to refer to 'this' host. Similarly a netid of all 0's means 'this' network (and the hostid field is ignored. All this is only valid in the appropriate context.

Also by convention an IP with a valid netid and a hostid of all 1's is a directed broadcast to all hosts on that net. An IP of all 1's is called a limited broadcast address or local network broadcast address and works independent of knowing the netid. Once a host learns netid it should use the limited broadcast.

In addition to unicast delivery where a packet is delivered to a single host, and broadcast delivery where a packet is delivered to all hosts on a given network, there is also a scheme called multicasting where a packet is delivered to a specific subset of hosts. These use Class D addresses.

Subnet and Supernet Extensions

In the 1980's when it became clear that not enough netids existed an extension called subnet addressing was developed to allow multiple physical networks to share a prefix. In the 1990's a second extension was specified that ignored the calssful hierarchy and allowed teh division between prefeix and suffix to occur at any point in the address. This is called classless addressing or supernetting, and allows greater utilization of the address space.

Hosts with multiple IP addresses because of connection to multiple networks (e.g. for routers) require that the path taken by packets traveling to such a host depends on the addressused.

Weaknesses in Internet Addressing

The disadvantage of encoding network information in an internat addres is that if a host moves from one network to another it must change its IP address. Another weakness is if a Class C network grows to have more than 254 hosts, it ust have its addresses changed to Class B. This must be done all at once for all hosts on the network and for all software that may have memorized IP addresses.

Dotted Decimal Notation

For communications with people, either in documents or with application programs, IP addresses are written as four decimal integers separated by decimal points, where each integere is the value of one octet or byte of the IP address starting with the most significant byte: e.g. 229.57.37.108

Loopback Address

A special address block in the class A range, 127.0.0.0, is reserved for loopback and is used for testing TCP/IP and for inter-process communications on the local computer. When any program uses any of the loopback addresses, the protocol software processes the data without actually sending it tp the network. The specifications require this. A router will never transmit such a packet if it somehwo does get on the net.

Reserved IP Addresses

IP Addresses, especially the netid must be unique a a net, whether it is the global internet or a private isolated network. For example, 9.0.0.0 has been assigned to IBm and 12.0.0.0 has been assigned to AT&T. These addresses can be used by an isolated intranet. The IETF has reserevd several address prefixes recommended for use on private networks.

Summary of Special IP Addresses

2005-2009