IPv6 defines several distinct address types — unicast, multicast, and anycast — each serving a specific communication role, while eliminating the broadcast model used in IPv4. Understanding these types and their assigned address ranges is fundamental to working with modern network infrastructure.
IPv6 Address Types
When it comes to address types, IPv6 did things that were quite a bit different than IP version 4. IPv4 had broadcast addresses, they had unicast addresses and multicast addresses. IPv6 has some of that, but they do things a little differently.
These are the address types that we have for IPv6. Now, one thing that you may notice is that there is no broadcast when it comes to IPv6. It's not listed out here because it doesn't exist with IPv6. IPv4 had a broadcast, where a machine would send out a broadcast and then that would be sent to all other machines on the network. IPv6 does have something similar, but it comes in the form of a multicast address, not a broadcast address. One of the things that IPv6 attempted to do was to get away from broadcasts on the network.
So what do we have? We have a unicast, which is just like IPv4, where you've got one machine that's talking to another machine and they're sending messages back and forth. You've got a multicast, which is just like IPv4, where you have one machine that's talking to multiple machines at once, but not all of the machines; it's not a broadcast. And then you have this new one called anycast.
Anycast is interesting. There's several machines that will subscribe to the anycast group, and then when a message is sent to that anycast address, what will happen is that message, that request, will only be sent to one of those machines, not all of those machines. See, multicast is one to many, but it's sending it to all of the machines, where anycast is going to send it to the first available machine that responds. So it's only going to be sent out to one of those machines and not all of the machines, and that's anycast.
With each one of these types there are other types as well. The ones that we're going to be going over are the globally unique address, a unique local address, and link local address. So just know that there are subtypes to these types.
A globally unique address is one that can be routed on the internet. So a device would get a globally unique address that would be good for traversing the network and going out to the rest of the world, that the rest of the world would know how to route back to that globally unique address.
Then we have a unique local address. A unique local address is within your own network here, so it stays within your own network, and so it's very much like an IP version 4 private address, except it's not globally addressable and you can't be translated. So it's never intended to go and be able to go out to the outside world with that address; it's only intended to be used internally.
Then we have our link local addresses. Our link local addresses are just supposed to be local; they're not supposed to be routed at all. What that means is that these are layer 3 devices, and this is a local area network right here. So a link local address is only meant to be used internally with this link. With this link local address, you're only going to be able to communicate within the devices that are in the layer 2 reach within your device.
IPv6 set out to do away with some things that IPv4 did. It wanted to do away with things like a DHCP server. It wanted you to be able to connect to a network, for a network interface to be able to determine its own IP address and start being able to communicate with its own IP address, and then be able to discover information to get out to the rest of the world. And so, in order to facilitate that, they came up with this link local address. As I mentioned, it can't be routed externally. It has to be configured on every interface, or it's actually automatically created: if it's an IPv6 interface, then it's going to automatically be created. And it's used to discover things on the network, so it'll start discovering ways of communicating on the network. And that's randomly generated. We'll talk about whether it's randomly generated or it could use EUI-64 to generate it, and we'll get more into that as well.
Here's a table that we're going to start filling out. We've been going over these different types of addresses, whether it be a globally unique address or a unique local address or a link local address. Well, they're assigned to certain ranges.
So we see that the globally unique address has things starting out with a 2000::/3. So the first three bits: if the first three bits is 0 1 0, then it's going to start everything from 2000 to 3FFF, and you can see all Fs there. So really it's a huge address space, a 2 to the 125th power. So it's a massive address space for those.
Unique local is also given a little bit smaller range, but still a really large range here. And the link local address is a little more zeroed in here; you've only got 2 to the 64th bit here of play room with this, but that's all we need, because that represents a ton amount of machines. So here are the ranges, and we're going to fill in the rest of this chart.
IPv6 definitely takes multicasting to the next level. Multicasting, remember, is when one device is talking to multiple devices on a network. There is one multicast address that I will mention, that is an all nodes. So it's not a broadcast, because it's not just sending it out to every device that's plugged into the network, but most devices are going to check in and subscribe to the all nodes multicast network. So most devices are actually going to receive messages from this all nodes multicast address. So there is that; that's IPv6's form of a broadcast, but it does work a little different than an actual broadcast. And we can see too that multicast has its own range as well, and it's quite a large range. So that is something that you'll see out there often, this FF00::/8.
Anycast is one of those unique aspects of IPv6. As mentioned before, it's much like multicast, where there's multiple devices that subscribe to the anycast server, but only one of them is going to get the request.
Let's say there's some sort of service that you are delivering and you want multiple devices to be able to deliver that service. Then multiple devices could possibly check in and say, hey, I'm available for requests going to this anycast address. So then when this machine makes a request to the anycast address, the closest machine, the one with the least cost, is going to be the one that ends up responding. So rather than it being sent out to all of these devices, it's only going to be sent out to the device that is closest to you, allowing us to free up quite a bit of compute space and bandwidth and things that are unnecessary. So we save on resources by these anycast addresses.
A few other ranges that I want to call out. This is by no means an exhaustive list of ranges; you can go and do a search online, look at a Wiki article, there's a lot more ranges. These are the ones that I just want to point out here.
We talked about the globally unique address, a unique local, a link local address, a multicast. The host loopback, this is similar to a 127.0.0.1, it's just ::1/128, so that's when it's referring to itself. Then we have an unspecified range, something that's not specified. And then we have the all route, or the default route, or the everything, that represents everything. This is similar to our quad zero when it comes to IPv4.
TechKnowSurge builds IT and cybersecurity professionals through hands-on, concept-first training built around real understanding — not memorization. Free interactive tools, structured programs, and 25+ years of real-world experience, all in one place.
Explore free tools and programs →