IPv4 Addresses and Subnets

An IP address is a unique 32-bit number assigned to a specific network node. Some IP addresses are more or less permanent, such as those assigned to vital servers; others change as required, such as those used by desktop clients. Individual machines on a shared network use IP addresses from a range of addresses assigned to that network.

Rather than expressing that 32-bit address as a single number, an IP address is divided into four 8-bit numbers, usually expressed as decimals. While 192.0.2.1 and 11000000.00000000.00000010.00000001 represent the same address, the first option is easier for our feeble little brains to grasp.

Internet service providers (ISPs) issue IP addresses in blocks. These blocks are the smallest allocation that they can get away with giving you—say, 16 or 32 addresses. If your system is on a server farm, you might get only a few IP addresses out of a block of 256.

A netmask indicates the size of the block of IP addresses assigned to your local network. The size of your IP block determines your netmask—or, your netmask determines how many IP addresses the network has.

ISPs issue IP addresses by prefix length, commonly called a slash. You’ll see IP address blocks described in forms like 192.0.2.128/26. Everyone who has worked with networking has seen the netmask 255.255.255.0, and most know that it’s associated with a block of 256 IP addresses. That netmask is also called a /24. The number after the slash is the number of fixed bits in the netmask. Remember, an IPv4 address is a 32-bit number; on a /24 network, 24 of those bits will never change.

This isn’t a textbook on binary math, so I won’t quiz you on the conversions, but think of an IP address as a string of 32 binary digits. On your networks, you can change the bits on the far right, but not the bits on the far left. But where is the line that separates right from left?

Netmasks have traditionally been split on 8-bit boundaries, but there’s no hard rule that says they must be. A /25 network has 25 fixed bits—one more fixed bit than what used to be called a class C network—leaving you with 7 bits to play with. The netmask’s fixed bits are set to 1, and your network bits are set to 0, as in the following example of a /25 netmask:

11111111.11111111.11111111.10000000

The first three blocks are set to the binary 11111111, which is 255 in decimal. The last block is set to 1000000, which is 128. Mash these together, and your resulting netmask is 255.255.255.128.

If you reduce netmasks to binary, they’re simple to figure out. While you won’t need to work with this every day, if you don’t understand the underlying concepts, the decimal conversion looks like total gibberish. With a little practice, you’ll recognize certain decimal strings as legitimate netmasks.

So now that you know how netmasks work, what the heck does all this mean in the real world?

IP addresses are issued in multiples of 2. If you have 4 bits to play with, you have 16 addresses (2⁴=16). If you have 8 bits to play with, you have 256 addresses (2⁸=256). If someone says that you have exactly 17 IP addresses, you’re either sharing a network with other people or they’re wrong.

It’s common to see a host’s IP with the netmask attached, such as 192.0.2.130/26. This gives you everything you need to attach the host to the local network. (Finding the default gateway is a separate issue, but it’s usually the top or bottom address in the block.)

$ ifconfig fxp0
…
        inet 192.0.2.226 netmask 0xfffffff0 broadcast 192.0.2.239
        inet 192.0.2.231 netmask 0xffffffff
…