0 TCP IP STEP 2 CLASSIFICATIONS

IP Addresses and Classifications
TCP/IP requires each host on a TCP/IP network to have its own unique IP address. An IP address is a 32-bit number, represented by a four-part decimal number (n.n.n.n), with each of the four parts - called octets - representing an 8-bit portion of the whole address number. Each octet can have a value ranging from 1 to 254. IP addresses used on the Internet are assigned and regulated by the Internet Assigned Numbers Authority, who in turn delegates the assignment process to the InterNIC. You should ask your Internet Service Provider for IP addresses when connecting your network to the Internet.

Another term to know is address pools. Address pools are blocks of IP addresses that are commonly referred to as networks. There are five types of address pools: Class A, Class B, Class C, Class D, and Class E. The first octet of the IP address represents the Class A network number; the second octet represents the Class B network number; and the third octet represents the Class C network number. The fourth octet of an IP address represents the actual host number. When you put all the octets together (e.g., 207.91.166.2), you have an IP address. Class D and Class E networks are special networks. They’re defined below, along with A, B, and C networks.
• Class A networks number large networks. The high-order bit of the first octet is always zero, which leaves seven bits in the first octet to define up to 127 networks. The remaining 24 bits of the 32-bit IP address are used to define the hosts, creating the possibility of 16,777,216 unique host addresses.
• Class B networks number medium-sized networks. The first two high-order bits of the first octet are always 10. Class B networks define up to 16,384 networks that can contain as many as 65,535 unique host addresses.
• Class C networks are typically used most often for numbering small networks. The first three bits are always 110, leaving room to define up to 2,097,152 networks, with each network having as many as 254 unique host addresses.
• Class D networks are made up of special multicast addresses that cannot be used for addressing devices on a network. The first four high-order bits are always set to 1110.
• Class E networks are reserved for experiments, with the first four high-order bits set to 1111.
IP Subnets
An IP subnet is a way of breaking a set of addresses, commonly referred to as an IP network, into smaller, more controllable pieces. When IP networks are subnetted, they can be routed independently to different gateways, which uses address space and bandwidth much more efficiently. An IP subnet operates by defining a subnet mask. The subnet mask is very similar in structure to an IP address in that it also has four parts, or octets. However, the allowable values of those octets are not quite the same as those in an IP address.

Subnet masks modify a 32-bit IP address by performing a mathematical calculation of the IP address and subnet mask at the bit level. Subnetting actually modifies the IP address by using the host address bits as additional network address bits.

The most common type of subnetting is performed on an even-byte boundary, although you may subnet a network using a bit boundary as well. An example of a subnet mask for a Class C network using an even-byte boundary is 255.255.255.0. If you use a Class C network address pool of 207.91.166.0, with usable host addresses ranging from 207.91.166.1 through 207.91.166.254, in combination with a subnet mask of 255.255.255.0, then all the addresses in that pool can be routed to the same gateway, and all are considered a part of the same logical IP network.

Subnetting can reduce the size of routing tables, minimize network traffic, isolate networks, maximize performance, and enhance your ability to secure a network. How you implement subnetting depends on many factors related to the use and design of your network.

Multihoming
A host may have more than one IP address; these hosts are known as multihomed hosts. Normally, hosts have just a single IP address; however, multihoming is necessary for several purposes, including establishing virtual Web servers. Multihoming can be useful for routing on a host computer containing more than one network card. Each card has its own unique IP address, and in practice, each of those cards could be connected to different physical network segments. With this type of configuration, more than one route to any given host can exist, which increases the probability of reaching it. This setup is useful when a network link goes down or becomes saturated with traffic.