Our TCP/IP networks are separated into many subnets. In this video, you’ll learn about the fundamentals of IP subnetting and how legacy class-based and modern classless subnetting is used to connect the worldwide Internet.
<< Previous Video: Network CablingNext: An Overview of IPv4 and IPv6 >>
When you first start getting into IP subnetting, one of the questions that comes up is why do we even have to go through this process at all? There’s a lot of math involved. We have to deal with converting between binary to decimal and back again. And it seems to be a very complex process. But the reality is we need to be able to subnets because of the complexities involved if we don’t.
Obviously we have so many devices, billions and billions of devices all over the world, that it would be very difficult if the device that was in our home would have to know the exact location of every single of these billions of devices all over the world. It’s just not possible for one device to be able to keep up and look up all of those different locations.
So what we have to do is split the network up into smaller pieces. This also makes it much more cost effective to have these devices on the network. And plus, they work a lot faster to be able to route our traffic from one place to the other.
Even very large devices tend to keep their subnets relatively small. They’ll generally have their subnets that their users are on, limited to about 253 devices or so. That makes it very, very easy then to be able to know exactly where someone might be and very easy to get traffic transferred from one place to the other using our very simple routers.
If you look at an IP address, it is actually more than just a single address. If we really start looking at an IP address, you’ll notice that it is always going to have on a device, both an IP address and a subnet mask. It is these two things that work together to help determine what network this particular device is on and what node on that network this particular device might be.
And if you know those two things, then you’ll also know if other devices might be on your same network based on their network address. And you also know if you need to go to a router to be able to talk to somebody who’s off of your subnet. So it’s really that combination of your IP address and your subnet mask that lets your workstation know exactly where it should be sending the data when it leaves your workstation.
There’s a historical concept with TCP/IP called a classful subnet. If we look at classful subnets, they are documented this way. Class A subnets are 255.0.0.0. Class B subnet masks would be 255.255.0.0. And Class C would be 255.255.255.0. And I’ve also shown you what the binary representation of those, right underneath the decimal subnet mask above this.
Now, we’ve are not used classful base subnetting since 1993. But we very commonly refer to it in common conversation. We might say that this particular subnet has a Class B subnet associated with it.
That’s much faster than saying that particular subnet uses a subnet mask of 255.255.0.0. We’ll just simply say that it’s a Class B subnet. It’s a shortcut to be able to describe to somebody else what the subnet mask might be.
The reason we have these class-based networks is we needed an automated way to determine what the subnet mask might be. It used to be we’d just assign an IP address and your workstation would already figure out what the subnet mask was based on the class of your IP address.
The way we would determine automatically what subnet to use is we would look at the first four bits of an IP address. If the first four bits started with a zero, that means that the IP addresses decimal number started between 1 and 126. And if we saw an IP address starting with those particular values, then we would automatically assign the default subnet mask of 255.0.0.0. That means that we would have 128 possible networks, with over 16 million hosts, on a Class A network.
If we were looking at for instance a Class C, that means that the first four bits of an IP address started with 110, which means that the value was between 192 and 223 decimal. And that means that we would automatically assign a mask of 255.255.255.0. And that would allow us to have over 2 million networks, with 254 hosts per network.
As you can see here, you’re very limited as to how many hosts and how many networks you can have. It’s all determined on what IP address you originally started with. That’s why we don’t use class-based subnetting any longer because it restricts us from having flexibility. And as the internet has grown, we’ve needed a lot more flexibility to get more granular subnetting so that we can have more subnets and more IP version 4 addresses on the internet.
That’s why in 1993, we got rid of the idea of using automated subnetting based on the IP address that you happen to be using. We instead started creating blocks that we could use to be able to differentiate between subnets. So this idea of Class A, and Class B, and Class C that is limited to certain IP addresses and has automatic default subnet masks, went completely out the window.
Instead, what we’ve done now is allow you to decide how many bits you’re going to use in your subnet mask. You might follow the old class-based standardizations. But now you have the flexibility to use different numbers of bits to be able to do your subnet masking.
For instance, you might have a 192.168.1.1 address that has a 24-bit subnet mask. Notice the /24 at the end. That’s very common to use in this CIDR block notation. The CIDR block notation /24 really means that the 24 bits in decimal are 255.255.255.0. And that means that your subnet is going to have this particular range.
But you can use any subnet that you would like. For instance this 10.1.0.1, that normally would be an 8-bit subnet if we had the old way of doing things in a class-based. In a classless configuration, we can make that a /16, which means the subnet mask is 255.255.0.0 and the address range would be this one.
This is usually the topic that confounds people. Because they were so used to memorizing that class-based chart, that they would see the 10 address, they would automatically determine that was a Class A. And therefore, the subnet mask always had to be 255.0.0.0.
But now, we don’t use that standardization. We give you more control on what you would like to make, so you could subnet it even further. And this is a good example of doing that.
Here’s another example of the same IP address, 10.1.0.1, with a 26-bit subnet. That means that the subnet mask is 255.255.255.192, which means that the range that you could have for IP addresses is 62 addresses per subnet.
Now obviously, there’s a lot of math and a lot of calculations that go on behind the scenes to calculate these. But the idea is that you now have flexibility on exactly how many networks you can configure and how many hosts per network.
Here’s an example of where you might use this classless, interdomain routing configuration. Here’s an example of an IP address that has being given out to what used to be freesoft.org. Their IP address used to be 18.104.22.168. And you’ll notice it has a 32-bit subnet mask. That means that it’s end station.
So how did we get to the point where that server was given that IP address? Well, we start with MCI, or what used to be MCI. They had an IP range, a big block of addresses that they could use, that was 22.214.171.124, with a subnet mask of /11. Again having that flexibility of that classless addressing allowed them not to use 8 bits of their subnet mask or 16 bits of that subnet mask.
They could fit something in that made more sense for what they were doing. And that was an 11-bit subnet mask. Well, one of the things that was provided to the Automation Research Systems Group from MCI was another block that was subnetted within that.
So MCI gave that network group a IP address range of 126.96.36.199, with a 22-bit subnet mask and said, you can do whatever you’d like to do in that 22-bit subnet mask. You could subnet it further.
And indeed they did. For their public servers, they created a subnet of 188.8.131.52, with a 24-bit subnet mask. And ultimately, they were able to put in that 24-bit subnet mask, a single server, which used to be freesoft.org.
That’s a very common scenario of taking a very, very long, big range of IP addresses, subnetting it down further, and allocating those subnets down to people further down the line. And ultimately, all the way down to a web server.