The TCP/IP Guide  9  TCP/IP Lower-Layer (Interface, Internet and Transport) Protocols (OSI Layers 2, 3 and 4)       9  TCP/IP Internet Layer (OSI Network Layer) Protocols            9  Internet Protocol (IP/IPv4, IPng/IPv6) and IP-Related Protocols (IP NAT, IPSec, Mobile IP)                 9  Internet Protocol Version 4 (IP, IPv4)                      9  IP Addressing                           9  IP Subnetting: Practical Subnet Design and Address Determination Example

IP Subnetting Step #1: Requirements Analysis 1 2 3 IP Subnetting Step #3: Determining The Custom Subnet Mask

After we complete our brief requirements analysis, we should know the two critical parameters that we must have in order to subnet our network: the number of subnets required for the network, and the maximum number of hosts per subnetwork. In using these figures to design our subnetted network, we will be faced with the key design decision in subnetting: how to divide the 8, 16 or 24 bits in the “classful” host ID into subnet ID and host ID.

Put another way, we need to decide how many bits to “steal” from the host ID to use for the subnet ID. As I introduced in the topic on custom subnet masks, the fundamental trade-off in choosing this number is as follows:

• Each bit taken from the host ID for the subnet ID doubles the number of subnets that are possible in the network.
  
  
• Each bit taken from the host ID for the subnet ID (approximately) halves the number of hosts that are possible within each subnet on the network.

There are six possible ways this decision can be made for a Class C network, as illustrated in Figure 73.

The relationship between the bits and the number of subnets and hosts is as follows:

• The number of subnets allowed in the network is two to the power of the number of subnet ID bits.
  
  
• The number of hosts allowed per subnet is two to the power of the number of host ID bits, less two.

We subtract two from the number of hosts in each subnet to exclude the “special meaning” cases where the host ID is all zeroes or all ones. As I explained in the topic on custom subnetting, this exclusion was originally also applied to the subnet ID, but is no longer in newer systems.

Now, to choose how many bits to use for the subnet we could use trial and error. By this I mean we could try to first calculate the number of subnets and hosts when we use one bit for the subnet ID and leave the rest for the host ID. We could then try with two bits for the subnet ID, and then try with three and so on. This would be silly, however; it's time consuming and makes it hard for us to choose the best option. There's an easier method: we can use the subnetting summary tables. They let us look at all our options and usually see immediately the best one for us.

IP Subnetting Step #1: Requirements Analysis 1 2 3 IP Subnetting Step #3: Determining The Custom Subnet Mask