IPv6 Overview


While the basic function of the IP is to move information across networks, IPv6 has more capabilities built into its foundation than IPv4. A key capability is the significant increase in address space. For example, all devices could have a public IP address so that they can be uniquely tracked.7 Today, inventory management of dispersed assets in a very large dispersed organization such as the United States Department of Defense (DoD) Department cannot be achieved with IP mechanisms; during the inventory cycle someone has to manually verify the location
of each desktop computer. With IPv6 one can use the network to verify that such equipment is there; even non-IT equipment in the field can also be tracked, by having an IP address permanently assigned to it. IPv6 also has extensive automatic configuration (autoconfiguration) mechanisms and reduces the IT burden, making configuration essentially plug-and-play (autoconfiguration implies that a Dynamic Host Configuration Protocol or DHCP server is not needed and/or does not have to be configured. Owing to the fact that IPv4 manual configuration is already a challenge in itself, one can understand that manually manipulating IPv6 addresses that are four times longer can be much more problematic. Corporations and government agencies will be able to achieve a number of improvements with IPv6 such as, but not limited to the following

  • expanded addressing capabilities;
  • serverless autoconfiguration (what some call “plug-n-play”) and reconfiguration;
  • streamlined header format and flow identification;
  • end-to-end security, with built-in, strong IP-layer encryption and authentication (embedded security support with mandatory IPsec implementation);
  • in IPv6, creating a VPN is easier and more standard than in IPv4, because of the Authentication Header (AH) and Encapsulating Security Protocol (ESP) Extension Headers and the performance penalty is lower for the VPN implemented in IPv6 compared to those built in IPv4 [25];
  • enhanced support for multicast and QoS (more refined support for flow control and QoS for the near real-time delivery of data);
  • more efficient and robust mobility mechanisms (enhanced support for Mobile IP and mobile computing devices);
  • extensibility: improved support for feature options/extensions;
  • IPv6 makes it easy for nodes to have multiple IPv6 addresses on the same network interface. This can create the opportunity for users to establish overlay or Communities of Interest (COI) networks on top of other physical IPv6 networks. Department, groups, or other users and resources can belong to one or more COIs, where each can have its own specific security policy [26];
  • merging two IPv4 networks with overlapping addresses (say, if two organizations merge) is complex; it will be much easier to merge networks with IPv6;
  • IPv6 network architectures can easily adapt to an end-to-end security model where the end hosts have the responsibility of providing the security services necessary to protect any data traffic between them; this results in greater flexibility for creating policy-based trust domains that are based on varying parameters including node address and application [27].

IPv6 basic capabilities include the following:

  • addressing,
  • anycast,
  • flow labels,
  • ICMPv6,
  • Neighbor Discovery (ND).

Table A5.1 shows the core protocols that comprise IPv6.

Key IPv6 ProtocolsIP was designed in the 1970s for the purpose of connecting computers that were in separate geographic locations. Computers in a campus were connected by means of local networks, but these local networks were separated into essentially stand-alone islands. “Internet,” as a name to designate the protocol and more recently the worldwide information network, simply means “internetwork”; that is, a connection between multiple networks. In the beginning, the protocol initially had only military use in mind, but computers from universities and enterprises were quickly added. The Internet as a worldwide information network is the result of the practical application of the IP protocol; that is, the result of the interconnection of a large set of information networks [19]. Starting in the early 1990s, developers realized that the communication needs of the twenty-first century required a protocol with some new features and capabilities, while at the
same time retaining the useful features of the existing protocol.

While link-level communication does not generally require a node identifier (address) since the device is intrinsically identified with the link-level address, communication over a group of links (a network) does require unique node identifiers (addresses). The IP address is an identifier that is applied to each device connected to an IP network. In this setup, different elements taking part in the network (servers, routers, desktop computers, etc.) communicate among each other using their IP address as an entity identifier. In version 4 of the IP protocol, addresses consist of four octets. For ease of human conversation, IP protocol addresses are represented as separated by periods, for example:, where the decimal numbers are a short hand (and correspond to) the binary code described by the byte in question (an 8 bit number takes a value in the 0–255 range). Since the IPv4 address has 32 bits there are nominally 232 different IP addresses (approximately 4 billion nodes, if all combinations are used). The Domain Name System (DNS) also helped the human conversation in the context of IPv4; DNS is going to be even more critical in IPv6 and will have substantial impact on security administrators that use IP addresses to define security policies (e.g., Firewalls).

IPv4 has proven, by means of its long life, to be a flexible and powerful networking mechanism. However, IPv4 is starting to exhibit limitations, not only with respect to the need for an increase of the IP address space, driven, for example, by new populations of users in countries such as China and India, and by new technologies with “always connected devices” (DSL, cable, networked Primary Deployment Area or PDAs, 2.5G/3G mobile telephones, etc.), but also in reference to a potential global rollout of VoIP. IPv6 creates a new IP address
format, so that the number of IP addresses will not get exhausted for several decades or longer even though an entirely new crop of devices are expected to connect to Internet.

IPv6 also adds improvements in areas such as routing and network autoconfiguration. Specifically, new devices that connect to Internet will be “plug-and-play” devices. With IPv6 one is not required to configure dynamic unpublished local IP addresses, the gateway address, the subnetwork mask or any other parameters. The equipment, when plugged into the network, automatically obtains all requisite
configuration data [19].

The advantages of IPv6 can be summarized as follows:

  • Scalability: IPv6 has 128 bit addresses versus 32 bit IPv4 addresses. With IPv4 the theoretical number of available IP addresses is 232 ∼ 1010. IPv6 offers a 2128 space. Hence, the number of available unique node addressees are 2128 ∼ 1039.
  • Security: IPv6 includes security features in its specifications such as payload encryption and authentication of the source of the communication.
  • Real-Time Applications: To provide better support for real-time traffic (e.g., VoIP), IPv6 includes “labeled flows” in its specifications. By means of this mechanism, routers can recognize the end-to-end flow to which transmitted packets belong. This is similar to the service offered by MPLS, but it is intrinsic with the IP mechanism rather than an add-on. Also, it preceded this MPLS feature by a number of years.
  • “Plug-And-Play”: IPv6 includes a “plug-and-play” mechanism that facilitates the connection of equipment to the network. The requisite configuration is automatic.
  • Mobility: IPv6 includes more efficient and enhanced mobility mechanisms, which are important for mobile networks.
  • Optimized Protocol: IPv6 embodies IPv4 best practices but removes unused or obsolete IPv4 characteristics. This results in a better-optimized Internet protocol.
  • Addressing and Routing: IPv6 improves the addressing and routing hierarchy.
  • Extensibility: IPv6 has been designed to be extensible and offers support for new options and extensions.

With IPv4, the 32-bit address can be represented as AdrClass|netID|hostID. The network portion can contain either a network ID or a network ID and a subnet. Every network and every host or device has a unique address, by definition. Basic NATing is a method by which IP addresses (specifically IPv4 addresses) are transparently mapped from one group to another. Specifically, private “unregistered”
addresses are mapped to a small set (as small as 1) of public registered addresses; this impacts the general addressability, accessibility, and “individuality” of the device. Network Address Port Translation (NAPT), also referred to as Port Address Translation (PAT), is a method by which many network addresses and their TCP/UDP ports are translated into a single network address and its TCP/UDP ports. Together, these two methods, referred to as traditional Network Address Translation (NAT), provide a mechanism to connect a realm with private
addresses to an external realm with globally unique registered addresses [29]. NAT is a short-term solution for the anticipated Internet growth requirements for this decade and a better solution is needed for address exhaustion. There is a clear recognition that NAT techniques make the Internet, the applications, and even the devices more complex (especially when conducting business-to-business transactions) and this means a cost overhead [19]. Overlapping encryptions domains has been a substantial issue for organizations to deal with when creating gateway-togateway VPNs. The expectation is that IPv6 can make IP devices less expensive, more powerful, and even consume less power; the power issue is not only important for environmental reasons, but also improves operability (e.g., longer battery
life in portable devices, such as mobile phones).

IPv4 addresses can be from an officially assigned public range or from an internal intranet private (but not globally unique) block. Internal intranet addresses may be in the ranges,, and, as suggested in RFC 1918. In the case of an internal intranet private address, a NAT function is employed to map the internal addresses to an external public address when the
private-to-public network boundary is crossed. This, however, imposes a number of limitations, particularly since the number of registered public addresses available to a company is almost invariably much smaller (as small as 1) than the number of internal devices requiring an address.

As noted, IPv4 theoretically allows up to 232 addresses, based on a four-octet address space. Public, globally unique addresses are assigned by the Internet Assigned Numbers Authority (IANA). IP addresses are addresses of network nodes at layer 3; each device on a network (whether the Internet or an intranet) must have a unique address. In IPv4, it is a 32-bit (4-byte) binary address used to identify the device. It is represented by the nomenclature a.b.c.d, each of a, b, c, and d being from 1 to 255 (0 has a special meaning). Examples are,, and

The problem is that during the 1980s many public, registered addresses were allocated to firms and organizations without any consistent control. As a result, some organizations have more addresses than they actually need, giving rise to the present dearth of available “registerable” Layer 3 addresses. Furthermore, not all IP addresses can be used due to the fragmentation described above.

One approach to the issue would be a renumbering and a reallocation of the IPv4 addressing space. However, this is not as simple as it appears since it requires significant worldwide coordination efforts and it would not solve the medium-term need for a much larger address space for evolving end-user/ consumer applications. Moreover, it would still be limited for the human population and the quantity of devices that will be connected to the Internet in the medium-term future [19]. At this juncture, and as a temporary and pragmatic approach to alleviate the dearth of addresses, NAT mechanisms are employed by organizations and even home users. This mechanism consists of using only a small set of public IPv4 addresses for an entire network to access to Internet. The myriad of internal devices are assigned IP addresses from a specifically designated range of Class A or Class C address that are locally unique but are duplicatively used and reused within various organizations. In some cases (e.g., residential Internet access use via DSL or cable), the legal IP address is only
provided to a user on a time-lease basis, rather than permanently.

A number of protocols cannot travel through a NAT device and hence the use of NAT implies that many applications (e.g., VoIP) cannot be used effectively in all instances.9 As a consequence, these applications can only be used in intranets. Examples include the following [19]:

  • Multimedia applications such as videoconferencing, VoIP, or VOD/IPTV do not work smoothly through NAT devices. Multimedia applications make use of RTP and Real-Time Control Protocol (RTCP). These in turn use UDP with dynamic allocation of ports and NAT does not directly support this environment.
  • IPsec is used extensively for data authentication, integrity, and confidentiality. However, when NAT is used, IPsec operation is impacted, since NAT changes the address in the IP header.
  • Multicast, although possible in theory, requires complex configuration in a NAT environment and hence, in practice, is not utilized as often as could be the case.

The need for obligatory use of NAT disappears with IPv6 (but it can still be used if someone wanted to).

The format of IPv6 addressing is described in RFC 2373. As noted, an IPv6 address consists of 128 bits, rather than 32 bits as with IPv4 addresses. The number of bits correlates to the address space, as follows:

NAT disappears with IPv6The relatively large size of the IPv6 address is designed to be subdivided into hierarchical routing domains that reflect the topology of the modern-day Internet. The use of 128 bits provides multiple levels of hierarchy and flexibility in designing hierarchical addressing and routing. The IPv4-based Internet currently lacks this flexibility [30].

The IPv6 address is represented as 8 groups of 16 bits each, separated by the “:” character. Each 16 bit group is represented by 4 hexadecimal digits, that is, each digit has a value between 0 and F (0,1, 2, . . . A, B, C, D, E, F with A = 1010, B = 1110, etc., to F = 1510). What follows is an example of a hypothetical IPv6 address

3223 : 0BA0:01E0:D001 : 0000 : 0000 : D0F0 : 0010

If one or more four-digit groups is 0000, the zeros may be omitted and replaced with two colons (::). For example,

3223 : 0BA0 ::

is the abbreviated form of the following address:

3223 : 0BA0 : 0000 : 0000 : 0000 : 0000 : 0000 : 0000

Similarly, only one 0 is written, removing 0’s in the left side, and four 0’s in the middle of the address. For example, the address

3223 : BA0 : 0 : 0 : 0 : 0 :: 1234

is the abbreviated form of the following address

3223 : 0BA0 : 0000 : 0000 : 0000 : 0000 : 0000 : 1234

There is also a method to designate groups of IP addresses or subnetworks that is based on specifying the number of bits that designate the subnetwork, beginning from left to right, using remaining bits to designate single devices inside the network. For example, the notation

3223 : 0BA0:01A0 :: /48

indicates that the part of the IP address used to represent the subnetwork has 48 bits. Since each hexadecimal digit has 4 bits, this points out that the part used to represent the subnetwork is formed by 12 digits, that is “3223:0BA0:01A0.” The remaining digits of the IP address would be used to represent nodes inside the network.

There are a number of special IPv6 addresses, as follows:

  • Autoreturn or Loopback Virtual Address: This address is specified in IPv4 as the address. In IPv6, this address is represented as ::1.
  • Unspecified Address (::): This address is not allocated to any node since it is used to indicate the absence of an address.
  • IPv6 over IPv4 Dynamic/Automatic Tunnel Addresses: These addresses are designated as IPv4-compatible IPv6 addresses and allow the sending of IPv6 traffic over IPv4 networks in a transparent manner. For example, they are represented as ::
  • IPv4 over IPv6 Addresses Automatic Representation: These addresses allow for IPv4-only-nodes to still work in IPv6 networks. They are designated as IPv4-mapped IPv6 addresses and are represented as ::FFFF: (e.g., ::FFFF:

Like IPv4, IPv6 is a connectionless, unreliable datagram protocol used primarily for addressing and routing packets between hosts. Connectionless means that a session is not established before exchanging data. Unreliable means that delivery is not guaranteed. IPv6 always makes a best-effort attempt to deliver a packet. An IPv6 packet might be lost, delivered out of sequence, duplicated, or delayed. IPv6 per se does not attempt to recover from these types of errors. The acknowledgment of packets delivered and the recovery of lost packets is done by a higher-layer protocol, such as TCP [30]. From a packet forwarding perspective, IPv6 operates just like IPv4.

An IPv6 packet, also known as an IPv6 datagram, consists of an IPv6 header and an IPv6 payload, as shown in Fig. A5.1. The IPv6 header consists of two parts, the IPv6 base header, and optional extension headers (Fig. A5.2). Functionally, the optional extension headers and upper-layer protocols, for example

IPv6 packet.IPv6 extension headers. IPv6 extension headers are optional headers that may follow the basic IPv6 header. An IPv6 PDU may include zero, one or multiple headers. When multiple extension headers are used, they form a chained list of headers identified by the “next header” field of the previous header.

TCP, are considered part of the IPv6 payload. Table A5.2 shows the fields in the IPv6 base header. IPv4 headers and IPv6 headers are not directly interoperable: hosts and/or routers must use an implementation of both IPv4 and IPv6 in order to recognize and process both header formats (Fig. A5.3). This gives rise to a number of complexities in the migration process between the IPv4 and the IPv6 environments. The IP header in IPv6 has been streamlined and defined to be of a fixed length (40 bytes). In IPv6, header fields from the IPv4 header have been removed, renamed, or moved to the new optional IPv6 Extension Headers. The header length field is no longer needed since the IPv6 header is now a fixed length entity. The IPv4 Type of Service is equivalent to the IPv6 Traffic Class field. The Total Length field has been replaced with the Payload Length field. Since IPv6 only allows for fragmentation to be performed by the IPv6 source
and destination nodes, and not individual routers, the IPv4 segment control fields (Identification, Flags, and Fragment Offset fields) have been moved to similar fields within the Fragment Extension Header. The functionality provided by the Time to Live (TTL10) field has been replaced with the Hop Limit field. The Protocol field has been replaced with the Next Header Type field. The Header Checksum field was removed; that has the main advantage of not having each relay spend time processing the checksum. The Options field is no longer part of

IPv6 Base HeaderComparison of IPv4 and IPv6 headersthe header as it was in IPv4. Options are specified in the optional IPv6 Extension Headers. The removal of the Options field from the header enables more efficient routing; only the information that is needed by a router needs to be processed [31].

One area requiring consideration, however, is the length of the IPv6 PDU: the 40-octet header can be a problem for real-time IP applications such as VoIP and IPTV. Header compression becomes critical [32].11 Also, there will be some bandwidth inefficiency in general, that could be an issue in limited-bandwidth environments or applications (e.g., sensor networks.)

“Autoconfiguration” is a new characteristic of the IPv6 protocol that facilitates network management and system setup tasks by users. This characteristic is often called “plug-and-play” or “connect-and-work.” Autoconfiguration facilitates initialization of user devices: after connecting a device to an IPv6 network, one or several IPv6 globally unique addresses are automatically allocated. DHCP allows systems to obtain an IPv4 address and other required information (e.g., default router or DNS server). A similar protocol, DHCPv6, has been published for IPv6. DHCP and DHCPv6 are known as stateful protocols because they maintain tables on (specialized) servers. However, IPv6 also has a new stateless autoconfiguration protocol that has no equivalent in IPv4. The stateless autoconfiguration protocol does not require a server component because there is no state to maintain (a DHCP server may typically run in a router or firewall). Every IPv6 system (other than routers) is able to build its own unicast global address. Stateless Address Autoconfiguration (SLAAC) provides an alternative between a purely manual configuration and stateful autoconfiguration [33].

“Stateless” autoconfiguration is also described as “serverless.” The acronym SLAAC is also used for serverless address autoconfiguration. SLAAC is defined in RFC 2462. With SLAAC, the presence of configuration servers to supply profile information is not required. The host generates its own address using a combination of the information that it possesses (in its interface or network card) and the information that is periodically supplied by the routers. Routers determine the prefix that identifies networks associated to the link under discussion. The “interface identifier” identifies an interface within a subnetwork and is often, and by default, generated from the Media Access Control (MAC) address of the network card. The IPv6 address is built combining the 64 bits of the interface identifier with the prefixes that routers determine as belonging to the subnetwork. If there is no router, the interface identifier is self-sufficient to allow the PC to generate a “link-local” address. The “link-local” address is sufficient to allow the communication between several nodes connected to the same link (the same local network).

IPv6 addresses are “leased” to an interface for a fixed established time (including an infinite time.) When this “lifetime” expires, the link between the interface and the address is invalidated and the address can be reallocated to other interfaces. For the suitable management of addresses expiration time, an address goes through two states (stages) while is affiliated to an interface [19]:

  1. At first, an address is in a “preferred” state, so its use in any communication is not restricted.
  2. After that, an address becomes “deprecated,” indicating that its affiliation with the current interface will (soon) be invalidated.

When it is in a “deprecated” state, the use of the address is discouraged, although it is not forbidden. However, when possible, any new communication (for example, the opening of a new TCP connection) must use a “preferred” address. A “deprecated” address should only be used by applications that have  already used it before and in cases where it is difficult to change this address to another address without causing a service interruption.

To ensure that allocated addresses (granted either by manual mechanisms or by autoconfiguration) are unique in a specific link, the link duplicated addresses detection algorithm is used. The address to which the duplicated address detection algorithm is being applied to is designated (until the end of this algorithmic session) as an “attempt address.” In this case, it does not matter that such an address has been allocated to an interface and received packets are discarded.

Next, we describe how an IPv6 address is formed. The lowest 64 bits of the address identify a specific interface and these bits are designated as “interface identifier.” The highest 64 bits of the address identify the “path” or the “prefix” of the network or router in one of the links to which such interface is connected. The IPv6 address is formed by combining the prefix with the interface identifier.

It is possible for a host or device to have IPv6 and IPv4 addresses simultaneously? Most of the systems that currently support IPv6 allow the simultaneous use of both protocols. In this way, it is possible to support communication with IPv4-only-networks as well as IPv6-only-networks and the use of the applications developed for both protocols [19].

Is it possible to transmit IPv6 traffic over IPv4 networks via tunneling methods. This approach consists of “wrapping” the IPv6 traffic as IPv4 payload data: IPv6 traffic is sent “encapsulated” into IPv4 traffic and at the receiving end, this traffic is parsed as IPv6 traffic. Transition mechanisms are methods used for the coexistence of IPv4 and/or IPv6 devices and networks. For example, an “IPv6-in- IPv4 tunnel” is a transition mechanism that allows IPv6 devices to communicate through an IPv4 network. The mechanism consists of creating the IPv6 packets in a normal way and encapsulating them in an IPv4 packet. The reverse process is undertaken in the destination machine that de-encapsulates the IPv6 packet.

There is a significant difference between the procedures to allocate IPv4 addresses, that focus on the parsimonious use of addresses (since addresses are a scare resource and should be managed with caution), and the procedures to allocate IPv6 addresses, that focus on flexibility. ISPs deploying IPv6 systems follow the RIRs policies relating to how to assign IPv6 addressing space among their clients. RIRs are recommending ISPs and operators allocate to each IPv6 client a/48 subnetwork; this allows clients to manage their own subnetworks without using NAT. (The implication is that the obligatory need for NAT disappears in IPv6).

In order to allow its maximum scalability, the IPv6 protocol uses an approach based on a basic header, with minimum information. This differentiates it from IPv4 where different options are included in addition to the basic header. IPv6 uses a header “concatenation” mechanism to support supplementary capabilities. The advantages of this approach include the following:

  • The size of the basic header is always the same, and is well known. The basic header has been simplified compared with IPv4, since only 8 fields are used instead of 12. The basic IPv6 header has a fixed size; hence, its processing by nodes and routers is more straightforward. Also, the header’s structure aligns to 64 bits, so that new and future processors (64 bits minimum) can process it in a more efficient way.
  • Routers placed between a source point and a destination point (that is, the route that a specific packet has to pass through), do not need to process or understand any “following headers.” In other words, in general, interior (core) points of the network (routers) only have to process the basic header while in IPv4, all headers must be processed. This flow mechanism is similar to the operation in MPLS, yet precedes it by several years.
  • There is no limit to the number of options that the headers can support (the IPv6 basic header is 40 octets in length, while IPv4 one varies from 20 to 60 octets, depending on the options used).

In IPv6, interior/core routers do not perform packets fragmentation, but the fragmentation is performed end-to-end. That is, source and destination nodes perform, by means of the IPv6 stack, the fragmentation of a packet and the reassembly, respectively. The fragmentation process consists of dividing the source packet into smaller packets or fragments [19].

The IPv6 specification defines a number of extension headers [31] (Table A5.3) [34]):

  • Routing Header: Similar to the source routing options in IPv4, the header is used to mandate a specific routing.
  • Authentication Header: AH is a security header that provides authentication and integrity.
  • Encapsulating Security Payload (ESP) Header: ESP is a security header that provides authentication and encryption.
  • Fragmentation Header: This is similar to the fragmentation options in IPv4. Destination Options Header: A header that contains a set of options to be processed only by the final destination node. Mobile IPv6 is an example of an environment that uses such a header
  • Hop-by-Hop Options Header: A set of options needed by routers to perform certain management or debugging functions..

As noted, IPsec provides network-level security where the application data is encapsulated within the IPv6 packet. IPsec utilizes the AH and/or ESP header to provide security (the AH and ESP header may be used separately or in combination). IPsec, with ESP, offers integrity and data origin authentication, confidentiality, and optional (at the discretion of the receiver) antireplay features (using confidentiality without integrity is discouraged by the RFCs); ESP furthermore provides limited traffic flow confidentiality. Both the AH and ESP header may be employed as follows [31] (Fig. A5.4):

IPv6 Extension Headers

  • Tunnel Mode: The protocol is applied to the entire IP packet. This method is needed to ensure security over the entire packet, where a new IPv6 header and an AH or ESP header are wrapped around the original IP packet.
  • Transport Mode: The protocol is just applied to the transport layer (i.e., TCP, UDP, ICMP) in the form of an IPv6 header, AH or ESP header, followed by the transport protocol data (header, data).

IPsec modes and types.

Migration to IPv6 environments is expected to be fairly complex. Initially, internetworking between the two environments will be critical. Existing IPv4- endpoints and/or nodes will need to run dual-stack nodes or convert to IPv6 systems. Fortunately, the new protocol supports an IPv4-compatible IPv6 address that is an IPv6 address employing embedded IPv4 addresses. Tunneling, that we already described in passing, will play a major role in the beginning. There are a number of requirements that are typically applicable to an organization wishing to introduce an IPv6 service [35]:

  • the existing IPv4 service should not be adversely disrupted (e.g., as it might be by router loading of encapsulating IPv6 in IPv4 for tunnels);
  • the IPv6 service should perform as well as the IPv4 service (e.g., at the IPv4 line rate, and with similar network characteristics);
  • the service must be manageable and be able to be monitored (thus tools should be available for IPv6 as they are for IPv4);
  • the security of the network should not be compromised, due to the additional protocol itself or a weakness of any transition mechanism used;
  • an IPv6 address allocation plan must be drawn up.

Well-known interworking mechanisms include the following [36]12:

  • Dual IP-Layer (or Dual Stack): A technique for providing complete support for both IPs—IPv4 and IPv6—in hosts and routers.
  • Configured Tunneling of IPv6 over IPv4: Point-to-point tunnels made by encapsulating IPv6 packets within IPv4 headers to carry them over IPv4 routing infrastructures.
  • Automatic Tunneling of IPv6 over IPv4: A mechanism for using IPv4- compatible addresses to automatically tunnel IPv6 packets over IPv4 networks.

Tunneling techniques include the following [36]12:

  • IPv6-over-IPv4 Tunneling: The technique of encapsulating IPv6 packets within IPv4 so that they can be carried across IPv4 routing infrastructures.
  • Configured Tunneling: IPv6-over-IPv4 tunneling where the IPv4 tunnel endpoint address is determined by configuration information on the encapsulating node. The tunnels can be either unidirectional or bidirectional. Bidirectional configured tunnels behave as virtual point-to-point links.
  • Automatic Tunneling: IPv6-over-IPv4 tunneling where the IPv4 tunnel endpoint address is determined from the IPv4 address embedded in the IPv4- compatible destination address of the IPv6 packet being tunneled.
  • IPv4 Multicast Tunneling: IPv6-over-IPv4 tunneling where the IPv4 tunnel endpoint address is determined using ND. Unlike configured tunneling, this does not require any address configuration and unlike automatic tunneling it does not require the use of IPv4-compatible addresses. However, the mechanism assumes that the IPv4 infrastructure supports IPv4 multicast.

Applications (and the lower-layer protocol stack) need to be properly equipped. There are four cases [37].

Case 1: IPv4-only applications in a dual-stack node. IPv6 protocol is introduced in a node, but applications are not yet ported to support IPv6. The protocol stack is as follows:

IPv6 protocol is introduced in a node

Case 2: IPv4-only applications and IPv6-only applications in a dual-stack node. Applications are ported for IPv6-only. Therefore there are two similar applications, one for each protocol version (e.g., ping and ping6). The protocol stack is as follows:

IPv4-only applications and IPv6-only applications

Case 3: Applications supporting both IPv4 and IPv6 in a dual-stack node. Applications are ported for both IPv4 and IPv6 support. Therefore, the existing IPv4 applications can be removed. The protocol stack is as follows:

Applications supporting both IPv4 and IPv6

Case 4: Applications supporting both IPv4 and IPv6 in an IPv4-only node. Applications are ported for both IPv4 and IPv6 support, but the same applications may also have to work when IPv6 is not being used (e.g., disabled from the OS). The protocol stack is as follows:

IPv4 and IPv6

The first two cases are not interesting in the longer term; only a few applications are inherently IPv4- or IPv6-specific and should work with both protocols without having to care about which one is being used.

Figure A5.5 depicts some basic scenarios of carrier-based IPv6 support. Cases (a) and (b) represent traditional environments where the carrier link supports either a clear channel that is used to connect, say, two IPv4 routers, or is IPaware. (In each case, the “cloud” on the left could also be the IPv4 Internet or the IPv6 Internet.)

Support of IPv6 in carrier networks.In Case (c), the carrier link is used to connect as a transparent link two IPv6 routers; the carrier link is not (does not need to be) aware that it is transferring IPv6 PDUs. In Case (d), the carrier system is IPv4-aware, so the use of that environment to support IPv6 requires IPv6 to operate in a tunneled-mode over the non-IPv6 cloud, which is a capability of IPv6.

In Case (e), the carrier infrastructure needs to provide a gateway function between the IPv4 and the IPv6 world (this could entail repacking the IP PDUs from the v4 format to the v6 format). Case (f) is the ideal long-term scenario where the “world has converted to IPv6” and “so did the carrier network.”

In Case (g), the carrier IP-aware network provides a conversion function to support both IPv4 (as a baseline) and IPv6 (as a “new technology”) handoffs. Possibly a dual-stack mechanism is utilized. In Case (h), the carrier IPv6-aware network provides a support function for IPv6 (as a baseline) and also a conversion function to support legacy IPv4 islands.

Even network/security administrators that operate in a pure IPv4 environment need to be aware of IPv6-related security issues. In a standard IPv4 environment where IPv6 is not explicitly supported, any form of IPv6-based tunneling traffic must be considered abnormal, malicious traffic. For example, unconstrained 6to4-based traffic should be blocked (6to4 is a transitional mechanism intended for individual independent nodes to connect IPv6 over the greater Internet). Most commercial-grade IPv4 firewalls block IP protocol 41, the 6to4, and tunnel protocol, unless it has been explicitly enabled [38].

In 2008, the Cooperative Association for Internet Data Analysis (CAIDA) and the American Registry for Internet Numbers (ARIN) surveyed over 200 respondents from USG agencies, commercial organizations (including ISPs and end users), educational institutions, associations, and other profit and nonprofit entities to determine the state of affairs in the United States with reference to IPv6 plans. Between 50% and 75% of the organizations surveyed indicated that they plan to deploy IPv6 by 2010 or sooner. According to some observers IPv6 is still an emerging technology, maturing and growing as practical experience is gained; others take a more aggressive view, as seen in the next section.