21.2 INTERNET ADDRESSING SCHEME

< Day Day Up >

Each end system on the network has to be uniquely identified. For this, the addressing scheme is very important. Since each end system is a node on a network, the addressing scheme should be such that the address contains both an ID for the network and an ID for the host. This scheme is followed in the IP addressing scheme. Each node on a TCP/IP network is identified by a 32-bit address. The address consists of the network ID and the host ID. IP address can be of five formats, as shown in Figure 21.1.

click to expand
Figure 21.1: IP address formats.

In IP Version 4, each system on the network is given a unique 32-bit IP address. The address consists of network ID and the host ID.

Class A addresses: Class A addressing is used when a site contains a small number of networks, and each network has many nodes (more than 65,536). Seven bits are used for network ID and 24 bits for host ID. A class A address has 0 in the first bit.

The maximum number of class A networks can be 126 (the network addresses 0 and 127 are reserved). Each network can accommodate up to (2²⁴ – 2) hosts. Note that two host addresses are reserved (all zeros and all ones).

The IP addresses are divided into five classes: A, B, C, D, and E. The number of bits assigned to the network ID field and the host ID field are different in each class.

Class B addresses: Class B addressing is used when a site has a medium number of networks and each network has more than 256 but less than 65,536 hosts. Fourteen bits are allocated for network ID and 16 bits for the host ID. A class B address has 10 for the first two bits.

Class C addresses: Class C addressing is used when a site has a large number of networks with each network having fewer than 256 hosts. Twenty-one bits are allocated to network ID and 8 bits to host ID. A class C address has 110 for the first three bits.

Class D addresses: These addresses are used when multicasting is required, such as when a datagram has to be sent to multiple hosts simultaneously.

Class E addresses: These addresses are reserved for future use.

In the IP address, if the host address bits are all zeros, the IP address represents the network address. If the host address bits are all ones, the IP address is the broadcast address—the packet is addressed to all hosts on the network.

It is possible that a host may be connected to different networks. A router is also connected to different networks. Such computers are known as multihomed hosts. These hosts need multiple IP addresses, each address corresponding to the machine's network connection. Hence, an IP address is given to the network connection.

When the sender wants to communicate over a network but does not know the network ID, network ID is set to all zeros. When that network sends a reply, it contains its network ID, which is recorded by the sender for future use.

Some of the drawbacks of this addressing scheme are:

In a class C network, if the number of hosts increases to more than 256, the whole addressing scheme has to be changed because network ID has to change. This calls for lots of work for the system administrator.
If a host is disconnected from one network and connected to another network, the IP address has to change.

Note

The addressing format of IP Version 4 is not an efficient addressing scheme because changing from one class of address to another class is very difficult.

An important point to be noted while transmitting the IP address is that integers are sent most significant byte first ("Big-Endian style") so that all the machines can interpret the correct address.

If a datagram has to be sent to multiple hosts, a multicast addressing scheme specified by class D addresses is used. IP multicast addresses can be assigned by a central authority (called well-known addresses) or temporarily created (called transient multicast groups). Multicast addressing is useful for applications such as audio and video conferencing.

In a class D addressing scheme, the first four bits are 1110. The remaining 28 bits identify the multicast address. Obviously, this address can be used only in the destination IP address and not in the source IP address.

The multicast address is used to send a datagram to multiple hosts. This addressing mechanism is required for applications such as audio/video conferencing.

An IP multicast address is mapped to the Ethernet multicast address by placing the lower 23 bits of the IP address into the low-order 23 bits of the Ethernet multicast address. However, note that this mapping is not unique, because five bits in the IP address are ignored. Hence, it is possible that some hosts on the Ethernet that are not part of the multicast group may receive a datagram erroneously, and it is the host's responsibility to discard the datagram.

Special routers called multicast routers are used to route datagrams with multicast addresses. Internet's multicast backbone (MBONE) has multicast routers that route the multicast traffic over the Internet. If a router does not support multicast routing, the mulitcast datagram is encapsulated in the normal unicast IP datagram, and the receiver has to interpret the multicast address.

21.2.1 Dotted Decimal Notation

Because it is difficult for us to read the IP address if it is written in 32-bit format, dotted decimal notation is used. If the IP address is

it can be represented as 254.127.129.170 for easy readability.

Internet Network Information Center (InterNIC) is the central authority to assign IP addresses. InterNIC gives the network ID and the organization is free to assign host addresses. Many organizations assign local IP addresses without obtaining the network ID from InterNIC. This is OK if the network remains isolated, but if it is connected to the Internet later on, there may be an address clash.

The 32-bit IP address is represented as a.b.c.d where the values of a, b, c and d can be between 0 and 255. This notation is called dotted decimal notation.

< Day Day Up >