I recently had to understand what CIDR blocks are and how they describe blocks of IP addresses and I figured I’d share what I learned.

I’ll stick to IPv4 addresses here, since that’s what most people are still dealing with these days.

Anatomy of an IP address

The Internet Protocol (IP) provides every device on a network with an IP address. They take the form a.b.c.d, where a, b, c and d are numbers from 0 to 255. Each of these four numbers is called an octet. It’s called an octet because it is a representation of 8 bits. If you’re not comfortable with how binary numbers work, you should stop here and take a quick tutorial.

Since an IP address has four octets and each octet has a value from 0-255, there are a total of 256 × 256 × 256 × 256 or 4,294,967,296 possible IP addresses (incidentally, the 4 billion limit is one of the reasons IPv6 came about).

Organisation of IP addresses

Although there are over 4 billion IP addresses available, they are not given out at random. There are specific IP ranges that are dedicated to specific purposes. For instance, you may have noticed that your home computer’s IP address is something like 192.168.0.5 or 192.168.1.20. That’s because the group of IP addresses that start 192.168 has been reserved for private networks. Another group that has been reserved for private use is the group that starts with a 10.

The Internet Assigned Numbers Authority (IANA) maintains the list of reserved IP addresses in a publicly available list. If you take a look you’ll notice that they use a special notation to define groups of IP address (or “CIDR Blocks”). The format is a.b.c.d/e. In this notation, e is the number that indicates the CIDR block and is called the prefix length.

Outside of the reserved blocks, CIDR blocks of different sizes are given out to organizations of different sizes. For instance, IBM has the 9.0.0.0/8 block. The UK Ministry of Defence has the 25.0.0.0/8 block.

CIDR blocks

The spectrum of IP address can be readily carved up into groups of IP addresses. Rather than create groups of arbitrary IP addresses, contiguous blocks are used; these blocks constitute networks within the Internet (the network of networks). As such, they are known as subnets. The CIDR notation allows us to concisely express a block of IP address that make up a subnet.

Subnets are identified by an IP address and the CIDR prefix length. An IP address is in a /e subnet if the first e bits of that IP address match the first e bits of the subnet address. Let’s see an example.

Suppose we have a subnet 192.168.1.0/24. Is the IP address 192.168.1.34 in this subnet?

First, convert the IP address to bits:

     192 .      168 .        1 .       34
11000000 . 10101000 . 00000001 . 00100010

Then take the first 24 bits, replace the remainder with zeros and convert them back to octets

11000000 . 10101000 . 00000001 . 00100010
11000000 . 10101000 . 00000001 . 00000000
     192 .      168 .        1 .        0

Here, we have our subnet address, so this IP address is in this subnet.

The prefix length tells us something about the size of the subnet. For a prefix length of n, we know that the last (32-n) bits of the IP address do not make a difference as to whether that IP address is in a given subnet. Thus, we can work out the size of the subnet.

CIDR prefix length Size of network Notes
32 1  
31 2  
30 4 For subnets this size and larger, the number of server IP addresses is usually reduced by 2. The lowest IP address is used as the network’s IP address and the highest IP address is used as the broadcast address.
29 8  
 
16 65,536 This is the size of the 192.168.x.x block, reserved for private networks
 
8 16,777,216 This is the size of the 10.x.x.x block, also reserved for private networks
 
0 4,294,967,296 This is the total set of IP addresses. Since many of the CIDR blocks are reserved, no practical network would have this prefix length.

Using a prefix on its own can be used when talking about non-specific CIDR blocks. Thus, we can talk about a /24 network and and know that we’re talking about a network with 4096 IP addresses in it.

CIDRs and subnet masks

If you’ve configured the network card of a computer before, you may recall that, along with the IP address of the computer, you sometime need to set the subnet mask. This is directly related to the CIDR prefix length. Essentially, the subnet mask for a prefix length n is the four octets you get when you set the first n bits to 1 and the remaining to 0. For example

CIDR prefix length Subnet mask
32 255.255.255.255
31 255.255.255.254
30 255.255.255.252
29 255.255.255.248
16 255.255.0.0
8 255.0.0.0
0 0.0.0.0

What are they for?

The main use of CIDR blocks is routing. When you request a web page from a server, your request needs to get out of your network, onto the Internet, into the network that contains the target webserver and to that actual server. In order to do this, it must go through several routers. These routers do not contain information about how to reach every single individual IP address on the internet. But the do contain information about how to get to different CIDR blocks.

Thus, your request can be passed through the internet to the appropriate CIDR block. From there, the router might know about other subnets within that CIDR block. For instance, 16.128.0.0/9 can, potentially, contain any number of subnets, so long as they are smaller and fit within that IP range. For instance, it could contain 16.128.0.64/26 or 16.128.32.0/19. By comparing the target IP address with the CIDR blocks that a router knows about, it can figure out where to send the packets next.

CIDR blocks can also be used by firewalls. CIDR blocks can be used to define whether packets to or from a specific block should be accepted or dropped.