The Historical Reasons for IPv6

In the last 40+ years, the Internet has gone from its infancy to being a huge influence in the world. It first grew through research at universities, from the ARPANET beginnings of the Internet in the late 1960s into the 1970s. The Internet kept growing fast in the 1980s, with the Internet’s fast growth still primarily driven by research and the universities that joined in that research. By the early 1990s, the Internet began to transform to allow commerce, allowing people to sell services and products over the Internet, which drove yet another steep spike upward in the growth of the Internet. Eventually, fixed Internet access (primarily through dial, digital subscriber line [DSL], and cable) became common, followed by the pervasive use of the Internet from mobile devices like smartphones. Figure 22-1 shows some of these major milestones with general dates.

The incredible growth of the Internet over a fairly long time created a big problem for public IPv4 addresses: the world was running out of addresses. For instance, in 2011, IANA allocated the final /8 address blocks (the same size as a Class A network), allocating one final /8 block to each of the five Regional Internet Registries (RIR). At that point, RIRs could no longer receive new allocations of public addresses from IANA to then turn around and assign smaller address blocks to companies or ISPs.

At that point in 2011, each of the five RIRs still had public addresses to allocate or assign. However, that same year, APNIC (Asia Pacific) became the first RIR to exhaust its available IPv4 address allocation. In late 2015, ARIN (North America) announced that it had exhausted its supply. When we were revising this chapter in 2019, IANA considered all RIRs except AFRINIC to have exhausted their supply of IPv4 addresses, with AFRINIC expected to run out of IPv4 address during the year 2019.

These events are significant in that the day has finally come in which new companies can attempt to connect to the Internet, but they can no longer simply use IPv4, ignoring IPv6. Their only option will be IPv6 because IPv4 has no public addresses left.

Even though the press has rightfully made a big deal about running out of IPv4 addresses, those who care about the Internet knew about this potential problem since the late 1980s. The problem, generally called the IPv4 address exhaustion problem, could literally have caused the huge growth of the Internet in the 1990s to have come to a screeching halt! Something had to be done.

The IETF came up with several short-term solutions to make IPv4 addresses last longer, and one long-term solution: IPv6. However, several other tools like Network Address Translation (NAT) and classless interdomain routing (CIDR) helped extend IPv4’s life another couple of decades. IPv6 creates a more permanent and long-lasting solution, replacing IPv4, with a new IPv6 header and new IPv6 addresses. The address size supports a huge number of addresses, solving the address shortage problem for generations (we hope). Figure 22-2 shows some of the major address exhaustion timing.

The rest of this first section examines IPv6, comparing it to IPv4, focusing on the common features of the two protocols. In particular, this section compares the protocols (including addresses), routing, routing protocols, and miscellaneous other related topics.

The IPv6 Protocols

The primary purpose of the core IPv6 protocol mirrors the same purpose of the IPv4 protocol. That core IPv6 protocol, as defined in RFC 2460, defines a packet concept, addresses for those packets, and the role of hosts and routers. These rules allow the devices to forward packets sourced by hosts, through multiple routers, so that they arrive at the correct destination host. (IPv4 defines those same concepts for IPv4 back in RFC 791.)

However, because IPv6 impacts so many other functions in a TCP/IP network, many more RFCs must define details of IPv6. Some other RFCs define how to migrate from IPv4 to IPv6. Others define new versions of familiar protocols or replace old protocols with new ones. For example:

Older OSPF Version 2 Upgraded to OSPF Version 3: The older Open Shortest Path First (OSPF) version 2 works for IPv4, but not for IPv6, so a newer version, OSPF version 3, was created to support IPv6. (Note: OSPFv3 was later upgraded to support advertising both IPv4 and IPv6 routes.)

ICMP Upgraded to ICMP Version 6: Internet Control Message Protocol (ICMP) worked well with IPv4 but needed to be changed to support IPv6. The new name is ICMPv6.

ARP Replaced by Neighbor Discovery Protocol: For IPv4, Address Resolution Protocol (ARP) discovers the MAC address used by neighbors. IPv6 replaces ARP with a more general Neighbor Discovery Protocol (NDP).

Although the term IPv6, when used broadly, includes many protocols, the one specific protocol called IPv6 defines the new 128-bit IPv6 address. Of course, writing these addresses in binary would be a problem—they probably would not even fit on the width of a piece of paper! IPv6 defines a shorter hexadecimal format, requiring at most 32 hexadecimal digits (one hex digit per 4 bits), with methods to abbreviate the hexadecimal addresses as well.

For example, all of the following are IPv6 addresses, each with 32 or fewer hex digits:

Click here to view code image

2345:1111:2222:3333:4444:5555:6666:AAAA
2000:1:2:3:4:5:6:A
FE80::1

The upcoming section “IPv6 Addressing Formats and Conventions” discusses the specifics of how to represent IPv6 addresses, including how to legally abbreviate the hex address values.

Like IPv4, IPv6 defines a header, with places to hold both the source and destination address fields. Compared to IPv4, the IPv6 header does make some other changes besides simply making the address fields larger. However, even though the IPv6 header is larger than an IPv4 header, the IPv6 header is actually simpler (on purpose), to reduce the work done each time a router must route an IPv6 packet. Figure 22-3 shows the required 40-byte part of the IPv6 header.

IPv6 Routing

As with many functions of IPv6, IPv6 routing looks just like IPv4 routing from a general perspective, with the differences being clear only once you look at the specifics. Keeping the discussion general for now, IPv6 uses these ideas the same way as IPv4:

• To be able to build and send IPv6 packets out an interface, end-user devices need an IPv6 address on that interface.

• End-user hosts need to know the IPv6 address of a default router, to which the host sends IPv6 packets if the host is in a different subnet.

• IPv6 routers de-encapsulate and re-encapsulate each IPv6 packet when routing the packet.

• IPv6 routers make routing decisions by comparing the IPv6 packet’s destination address to the router’s IPv6 routing table; the matched route lists directions of where to send the IPv6 packet next.

While the list shows some concepts that should be familiar from IPv4, the next few figures show the concepts with an example. First, Figure 22-4 shows a few settings on a host.

The host (PC1) has an address of 2345::1. PC1 also knows its default gateway of 2345::2. (Both values are valid abbreviations for real IPv6 addresses.) To send an IPv6 packet to host PC2, on another IPv6 subnet, PC1 creates an IPv6 packet and sends it to R1, PC1’s default gateway.

The router (R1) has many small tasks to do when forwarding this IPv6 packet, but for now, focus on the work R1 does related to encapsulation. As seen in Step 1 of Figure 22-5, R1 receives the incoming data-link frame and extracts (de-encapsulates) the IPv6 packet from inside the frame, discarding the original data-link header and trailer. At Step 2, once R1 knows to forward the IPv6 packet to R2, R1 adds a correct outgoing data-link header and trailer to the IPv6 packet, encapsulating the IPv6 packet.

When a router like R1 de-encapsulates the packet from the data-link frame, it must also decide what type of packet sits inside the frame. To do so, the router must look at a protocol type field in the data-link header, which identifies the type of packet inside the data-link frame. Today, most data-link frames carry either an IPv4 packet or an IPv6 packet.

To route an IPv6 packet, a router must use its IPv6 routing table instead of the IPv4 routing table. The router must look at the packet’s destination IPv6 address and compare that address to the router’s current IPv6 routing table. The router uses the forwarding instructions in the matched IPv6 route to forward the IPv6 packet. Figure 22-6 shows the overall process.

Note that again, the process works like IPv4, except that the IPv6 packet lists IPv6 addresses, and the IPv6 routing table lists routing information for IPv6 subnets (called prefixes).

Finally, in most enterprise networks, the routers will route both IPv4 and IPv6 packets at the same time. That is, your company will not decide to adopt IPv6, and then late one weekend night turn off all IPv4 and enable IPv6 on every device. Instead, IPv6 allows for a slow migration, during which some or all routers forward both IPv4 and IPv6 packets. (The migration strategy of running both IPv4 and IPv6 is called dual stack.) All you have to do is configure the router to route IPv6 packets, in addition to the existing configuration for routing IPv4 packets.

IPv6 Routing Protocols

IPv6 routers need to learn routes for all the possible IPv6 prefixes (subnets). Just like with IPv4, IPv6 routers use routing protocols, with familiar names, and generally speaking, with familiar functions.

None of the IPv4 routing protocols could be used to advertise IPv6 routes originally. They all required some kind of update to add messages, protocols, and rules to support IPv6. Over time, Routing Information Protocol (RIP), Open Shortest Path First (OSPF), Enhanced Interior Gateway Routing Protocol (EIGRP), and Border Gateway Protocol (BGP) were all updated to support IPv6. Table 22-2 lists the names of these routing protocols, with a few comments.

In addition, these routing protocols also follow the same interior gateway protocol (IGP) and exterior gateway protocol (EGP) conventions as their IPv4 cousins. RIPng, EIGRPv6, and OSPFv3 act as interior gateway protocols, advertising IPv6 routes inside an enterprise.

As you can see from this introduction, IPv6 uses many of the same big ideas as IPv4. Both define headers with a source and destination address. Both define the routing of packets, with the routing process discarding old data-link headers and trailers when forwarding the packets. And routers use the same general process to make a routing decision, comparing the packet’s destination IP address to the routing table.

The big differences between IPv4 and IPv6 revolve around the bigger IPv6 addresses. The next topic begins looking at the specifics of these IPv6 addresses.

Representing Full (Unabbreviated) IPv6 Addresses

IPv6 uses a convenient hexadecimal (hex) format for addresses. To make it more readable, IPv6 uses a format with eight sets of four hex digits, with each set of four digits separated by a colon. For example:

Click here to view code image

2340:1111:AAAA:0001:1234:5678:9ABC:1234

IPv6 addresses also have a binary format as well, but thankfully, most of the time you do not need to look at the binary version of the addresses. However, in those cases, converting from hex to binary is relatively easy. Just change each hex digit to the equivalent 4-bit value listed in Table 22-3.

Abbreviating and Expanding IPv6 Addresses

IPv6 also defines ways to abbreviate or shorten how you write or type an IPv6 address. Why? Although using a 32-digit hex number works much better than working with a 128-bit binary number, 32 hex digits are still a lot of digits to remember, recognize in command output, and type on a command line. The IPv6 address abbreviation rules let you shorten these numbers.

Computers and routers typically use the shortest abbreviation, even if you type all 32 hex digits of the address. So even if you would prefer to use the longer unabbreviated version of the IPv6 address, you need to be ready to interpret the meaning of an abbreviated IPv6 address as listed by a router or host. This section first looks at abbreviating addresses and then at expanding addresses.

Abbreviating IPv6 Addresses

Two basic rules let you, or any computer, shorten or abbreviate an IPv6 address:

1. Inside each quartet of four hex digits, remove the leading 0s (0s on the left side of the quartet) in the three positions on the left. (Note: at this step, a quartet of 0000 will leave a single 0.)

2. Find any string of two or more consecutive quartets of all hex 0s, and replace that set of quartets with a double colon (::). The :: means “two or more quartets of all 0s.” However, you can use :: only once in a single address because otherwise the exact IPv6 might not be clear.

For example, consider the following IPv6 address. The bold digits represent digits in which the address could be abbreviated.

Click here to view code image

FE00:0000:0000:0001:0000:0000:0000:0056

Applying the first rule, you would look at all eight quartets independently. In each, remove all the leading 0s. Note that five of the quartets have four 0s, so for these, remove only three 0s, leaving the following value:

FE00:0:0:1:0:0:0:56

While this abbreviation is valid, the address can be abbreviated more, using the second rule. In this case, two instances exist where more than one quartet in a row has only a 0. Pick the longest such sequence, and replace it with ::, giving you the shortest legal abbreviation:

FE00:0:0:1::56

While FE00:0:0:1::56 is indeed the shortest abbreviation, this example happens to make it easier to see the two most common mistakes when abbreviating IPv6 addresses. First, never remove trailing 0s in a quartet (0s on the right side of the quartet). In this case, the first quartet of FE00 cannot be shortened at all because the two 0s trail. So, the following address, which begins now with only FE in the first quartet, is not a correct abbreviation of the original IPv6 address:

FE:0:0:1::56

The second common mistake is to replace all series of all 0 quartets with a double colon. For example, the following abbreviation would be incorrect for the original IPv6 address listed in this topic:

FE00::1::56

The reason this abbreviation is incorrect is that now you do not know how many quartets of all 0s to substitute into each :: to find the original unabbreviated address.

Expanding Abbreviated IPv6 Addresses

To expand an IPv6 address back into its full unabbreviated 32-digit number, use two similar rules. The rules basically reverse the logic of the previous two rules:

1. In each quartet, add leading 0s as needed until the quartet has four hex digits.

2. If a double colon (::) exists, count the quartets currently shown; the total should be less than 8. Replace the :: with multiple quartets of 0000 so that eight total quartets exist.

The best way to get comfortable with these addresses and abbreviations is to do some yourself. Table 22-4 lists some practice problems, with the full 32-digit IPv6 address on the left and the best abbreviation on the right. The table gives you either the expanded or abbreviated address, and you need to supply the opposite value. The answers sit at the end of the chapter, in the section “Answers to Earlier Practice Problems.”

Full	Abbreviation
2340:0000:0010:0100:1000:ABCD:0101:1010
	30A0:ABCD:EF12:3456:ABC:B0B0:9999:9009
2222:3333:4444:5555:0000:0000:6060:0707
	3210::
210F:0000:0000:0000:CCCC:0000:0000:000D
	34BA:B:B::20
FE80:0000:0000:0000:DEAD:BEFF:FEEF:CAFE
	FE80::FACE:BAFF:FEBE:CAFE

Representing the Prefix Length of an Address

IPv6 uses a mask concept, called the prefix length, similar to IPv4 subnet masks. Similar to the IPv4 prefix-style mask, the IPv6 prefix length is written as a /, followed by a decimal number. The prefix length defines how many bits of the IPv6 address define the IPv6 prefix, which is basically the same concept as the IPv4 subnet ID.

When writing an IPv6 address and prefix length in documentation, you can choose to leave a space before the /, or not, as shown in the next two examples.

2222:1111:0:1:A:B:C:D/64
2222:1111:0:1:A:B:C:D /64

Finally, note that the prefix length is a number of bits, so with IPv6, the legal value range is from 0 through 128, inclusive.

Calculating the IPv6 Prefix (Subnet ID)

With IPv4, you can take an IP address and the associated subnet mask, and calculate the subnet ID. With IPv6 subnetting, you can take an IPv6 address and the associated prefix length, and calculate the IPv6 equivalent of the subnet ID: an IPv6 prefix.

Like with different IPv4 subnet masks, some IPv6 prefix lengths make for an easy math problem to find the IPv6 prefix, while some prefix lengths make the math more difficult. This section looks at the easier cases, mainly because the size of the IPv6 address space lets us all choose to use IPv6 prefix lengths that make the math much easier.

Finding the IPv6 Prefix

In IPv6, a prefix represents a group of IPv6 addresses. For now, this section focuses on the math, and only the math, for finding the number that represents that prefix. Chapter 23, “IPv6 Addressing and Subnetting,” then starts putting more meaning behind the actual numbers.

Each IPv6 prefix, or subnet if you prefer, has a number that represents the group. Per the IPv6 RFCs, the number itself is also called the prefix, but many people just call it a subnet number or subnet ID, using the same terms as IPv4.

As with IPv4, you can start with an IPv6 address and prefix length, and find the prefix, with the same general rules that you use in IPv4. If the prefix length is /P, use these rules:

1. Copy the first P bits.

2. Change the rest of the bits to 0.

When using a prefix length that happens to be a multiple of 4, you do not have to think in terms of bits, but in terms of hex digits. A prefix length that is a multiple of 4 means that each hex digit is either copied or changed to hex 0. Just for completeness, if the prefix length is indeed a multiple of 4, the process becomes

1. Identify the number of hex digits in the prefix by dividing the prefix length (which is in bits) by 4.

2. Copy the hex digits determined to be in the prefix per the first step.

3. Change the rest of the hex digits to 0.

Figure 22-7 shows an example, with a prefix length of 64. In this case, Step 1 looks at the /64 prefix length and calculates that the prefix has 16 hex digits. Step 2 copies the first 16 digits of the IPv6 address, while Step 3 records hex 0s for the rest of the digits.

After you find the IPv6 prefix, you should also be ready to abbreviate the IPv6 prefix using the same rules you use to abbreviate IPv6 addresses. However, you should pay extra attention to the end of the prefix because it often has several octets of all 0 values. As a result, the abbreviation typically ends with two colons (::).

For example, consider the following IPv6 address that is assigned to a host on a LAN:

Click here to view code image

2000:1234:5678:9ABC:1234:5678:9ABC:1111/64

This example shows an IPv6 address that itself cannot be abbreviated. After you calculate the prefix for the subnet in which the address resides, by zeroing out the last 64 bits (16 digits) of the address, you find the following prefix value:

Click here to view code image

2000:1234:5678:9ABC:0000:0000:0000:0000/64

This value can be abbreviated, with four quartets of all 0s at the end, as follows:

2000:1234:5678:9ABC::/64

To get better at the math, take some time to work through finding the prefix for several practice problems, as listed in Table 22-5. The answers sit at the end of the chapter, in the section “Answers to Earlier Practice Problems.”

Working with More-Difficult IPv6 Prefix Lengths

Some prefix lengths make the math to find the prefix very easy, some mostly easy, and some require you to work in binary. If the prefix length is a multiple of 16, the process of copying part of the address copies entire quartets. If the prefix length is not a multiple of 16 but is a multiple of 4, at least the boundary sits at the edge of a hex digit, so you can avoid working in binary.

Although the /64 prefix length is by far the most common prefix length, you should be ready to find the prefix when using a prefix length that is any multiple of 4. For example, consider the following IPv6 address and prefix length:

Click here to view code image

2000:1234:5678:9ABC:1234:5678:9ABC:1111/56

Because this example uses a /56 prefix length, the prefix includes the first 56 bits, or first 14 complete hex digits, of the address. The rest of the hex digits will be 0, resulting in the following prefix:

Click here to view code image

2000:1234:5678:9A00:0000:0000:0000:0000/56

This value can be abbreviated, with four quartets of all 0s at the end, as follows:

2000:1234:5678:9A00::/56

This example shows an easy place to make a mistake. Sometimes, people look at the /56 and think of that as the first 14 hex digits, which is correct. However, they then copy the first 14 hex digits and add a double colon, showing the following:

2000:1234:5678:9A::/56

This abbreviation is not correct because it removed the trailing “00” at the end of the fourth quartet. If you expanded the abbreviated value, it would begin with 2000:1234:5678:009A, not 2000:1234:5678:9A00. So, be careful when abbreviating when the boundary is not at the edge of a quartet.

Once again, some extra practice can help. Table 22-6 uses examples that have a prefix length that is a multiple of 4, but is not on a quartet boundary, just to get some extra practice. The answers sit at the end of the chapter, in the section “Answers to Earlier Practice Problems.”