Addressing the Shortage

IP version 6 should solve the critical IP address shortage pending with IP version 4.

As wireless carriers look to the future, one technology that looms large is IP. Although wireless carriers have traditionally relied on circuit-switched networks to handle predominantly mobile voice traffic, the projected growth of wireless data traffic from 2% today to 23% in 2002 will demand architectural changes. Carriers will need to transform their networks not only for packet data but also to handle the increased traffic more efficiently.

IP is the ideal platform for moving carriers from traditional circuit-switched network architectures to next-generation, IP-optimized, packet-based networks. As the worldwide standard of networking, IP offers many benefits, including compatibility with existing networks, lower network operational costs and greater speed in bringing services to market.

First developed in 1976, IP version 4 (IPv4) has given birth to some of the most significant technology developments ever, most notably the Internet. This technology has been the protocol that has scaled with the Internet for the past 25 years. Although IP is clearly the future of wireless, wireless applications present new challenges that will not be solved easily with this existing standard. Although IPv4 still has a very important function as a protocol for the Internet, a pending address shortage has prompted the Internet Engineering Task Force (IETF) (www.ietf.org) to develop the IP version 6 (IPv6) protocol.

No. 1 Challenge: Addresses

One of the most profound challenges facing the Internet community in general and the wireless industry in particular is that of addresses. Although IPv4 theoretically can support as many as 4 billion unique addresses, the actual allocation of space has locked up nearly 75% of these addresses. In the early days, a limited number of entities, such as MIT and AT&T, were each allocated Class-A IP networks. Today, each controls more than 16 million addresses. Consequently, companies that today apply for IP addresses must make do with a fraction of the remaining Class B and C addresses.

On top of this, Internet traffic still is growing at 400% a year worldwide 320 million users were connected in 2000, and 550 million are projected by 2005. Geopolitical issues are emerging because countries such as China, India, Japan and Korea, for example, only have a small percentage of the total available IP address space.

By selecting IPv6 as their protocol of choice, 3G Partnership Project (3GPP) (www.3gpp.org) and Universal Mobile Telecommunication System (UMTS) (www.umts-forum.org) organizations have built a strong business case to move mobile users toward IPv6. In 2000, 400 million people were reachable on mobile phones; more than one billion are expected to go mobile by 2005. Of the one billion cars that will be produced in 2010, at least 15% are expected to be equipped with mobile Internet connectivity such as GPS and yellow-page services. What's more, billions of new always-on Internet appliances for the home will be connected through various technologies, and each device will require its own IP address.

In the mobile arena, the issue is exacerbated by the fact that mobile devices need to use an official global IP address at each new point of attachment to the Internet when mobile IP is not used. They have to be reachable in the same way you connect to a Web server today, and with IPv4, getting these addresses is not always easy.

The anticipated roll-out of wireless data services has been identified as a key IPv6 driver, and this is reflected in the fact that the key industry organizations Mobile Wireless Internet Forum (www.mwif.org), 3GPP and UMTS have selected IPv6 as the foundation for future IP services.

Address Abundance

Among the many benefits that IPv6 will deliver are the abundance of addresses ISPs will have to allocate to the many mobile users and Internet-capable devices that will rapidly proliferate in the coming years. The availability of globally unique addresses for these devices can eliminate the need for Network Address Translation (NAT), which is currently used to extend the available IPv4 address capacity. The elimination of NAT also increases security by enabling end-to-end encryption and transmission of digital signatures. Using an example from crowded telephone networks, one might say that IPv6 will eliminate the need for extensions, giving every office direct communication lines without requiring operators (automatic or otherwise) to redirect calls.

In contrast to IPv4, which has 32 bits of address space, IPv6 has 128 bits of address space, pushing the theoretical limit of unique IPv6 nodes to roughly 3.4 10³⁸ or about 340 billion billion billion billion unique addresses, according to Latif Ladid, president of the IPv6 Forum a major advocacy group for IPv6 (www.ipv6forum.com) and Ericsson Telebit (www.tbit.dk) vice president.

What About NAT?

Some believe that techniques such as NAT, which has been used to extend the number of available IPv4 addresses, can be modified to keep IPv4 solvent for a few more years. NAT does this by using a single IP address for an entire network and uses a converter to create distinct internal addresses for each device on the network. NAT was designed with a client-server model where servers have an official address, clients being behind a NAT device. This model cannot apply to the new always-on devices, which look like servers from an IP service view.

Opponents of NAT believe that it is an imperfect workaround solution that is harmful to interoperability because it disables end-to-end networking features a key capability for security. Because every datagram that passes across a NAT network must be converted, NAT destroys the ability to use IP security architecture (IPsec) protocols to encrypt or sign transmissions digitally. Because IPv6 can eliminate the need for NAT by providing as many real IP addresses as organizations need, IPv6 supports full end-to-end encryption or digital signatures.

Other Benefits of IPv6

Beyond addresses and security, other advantages of IPv6 over IPv4 include:

• Plug and play. Overall, the cost of running an IPv6 network to serve the new Internet appliances will be less than that of running an equivalent IPv4 network. This is because IPv6 integrates plug-and-play features not available on IPv4. For example, IPv6 nodes can automatically configure their addresses themselves without the need of any server such as IPv4 DHCP, although this feature still exists for network managers who do not want to use the auto-configuration mechanism. These features make for true plug-and-play network access. Through neighbor discovery, nodes can automatically determine which routers on their links are available and can be reached. Moreover, the assignment of IP addresses is simplified at the organizational level.

• Mobile IP. One of the key IPv6 features is enhanced mobility. Because mobility is not built into the IPv4 protocol, network designers have had to build a solution called triangle routing to support mobile IP. But triangle routing is inefficient for a large-scale deployment.

In triangle routing, each time a mobile host wishes to send data, it sends it to the home network of the mobile host, which then communicates with the current location of the mobile host. The need to send each packet across this distance may be highly inefficient and creates greater latency. Because of the extra steps required to set up communications, it makes it more difficult for the mobile host to ask for the right resources at the right time when the data sent are multimedia types.

IPv6, on the other hand, includes built-in mechanisms that allow a host to be informed of the temporary address of the mobile host and, thereby, to send the packets directly to the mobile host. This means the traffic must travel less distance, reducing network congestion and improving latency to offer better performance.

Managing the Transition

As the transition to IPv6 moves forward, its ultimate success will depend on the ability of the industry to integrate it within today's existing IPv4 infrastructure without disruption of services. It's widely believed that as the transition occurs, both protocols will need to coexist for an indefinite period of time.

Many techniques have been identified and are being implemented for managing the transition. They basically fall into three broad categories:

• Dual-stack techniques to allow IPv4 and IPv6 to coexist in the same devices and networks

• Tunneling techniques to allow IPv6 devices at the edges to communicate over an IPv4 backbone

• IPv6-over-dedicated-data-link techniques to allow IPv6 devices to communicate over a separate IPv6 infrastructure.

Dual-Stack Approach

In the dual-stack approach, host devices will run both protocols simultaneously, enabling applications to migrate one at a time to an IPv6 transport. This approach will be useful mainly for applications communicating with both IPv4 and IPv6 devices. Supporting dual stacks is similar to the support of multiprotocol routing such as AppleTalk, IPX, DECnet, and others in the late 1980s. In most cases, IPv6 will be bundled with new OS releases and will not be an extra-cost add-on.

The dual-stack approach will allow indefinite coexistence of IPv4 and IPv6 and will enable gradual application-by-application upgrades to the emerging IPv6 standard. (See Figure 1 on page 52.)

Tunneling

Tunneling techniques are being developed for use in those environments in which IPv6 networks must communicate across an IPv4 transit network. In tunneling, IPv6 traffic is encapsulated within IPv4 packets or multiprotocol label-switching (MPLS) frames for the duration of the transit path across the IPv4 network.

In Figure 2 on page 56, IPv6 enterprises or IPv6 mobile users communicate across an IPv4 or MPLS infrastructure through tunneling. An encapsulating gateway at the edge of the network sets up the tunnels to carry the IPv6 data to their destination. This will have no impact on the existing IPv4 or MPLS-based backbone and will be enabled using high-speed packet-over-SONET (PoS) interfaces.

The advantage of this approach is that it starts at the edge which is only required when IPv6 communications are needed. The Edge ingress and egress router upgrades until native IPv6 networks are offered end-to-end. This approach also can be viewed as an IPv6 virtual public network running over the IPv4 Internet and which will become less virtual over time.

IPv6 Integration With MPLS

MPLS, a forwarding-and-control plane architecture pioneered by Cisco (www.cisco.com) and standardized by the IETF, is an architectural improvement that runs on top of the underlying Internet infrastructure. Carriers are deploying MPLS because of the compelling benefits in terms of traffic engineering, class-of-service transparency and virtual private network (VPN) functionality.

Many believe that MPLS and IPv6 are highly complementary and that integrating IPv6 Edge routers over an MPLS topology requires much less backbone infrastructure upgrading or reconfiguration. This is because MPLS forwarding is based upon labels rather than the IP header itself, eliminating the need for network core hardware and software to be upgraded a likely reality for native end-to-end IPv6 forwarding.

Additionally, the inherent VPN and traffic-engineering services available within an MPLS environment could enable IPv6 networks going forward to be combined into VPNs or extranets over an infrastructure also supporting IPv4 VPNs and MPLS.

Considering the evolution of a wireless infrastructure to 3GPP/UMTS, a wireless carrier can deploy a MPLS network today, supporting circuit-switching and data switching as required in the Release 97 & 99 specifications, then add IPv6 services by configuring Edge routers that have IPv6 enabled on its MPLS infrastructure, avoiding any major network changes.

IPv6 Over Dedicated Data Links

For several years, many carriers have designed WANs or MANs for data services, deploying Layer 2 technologies such as frame relay and ATM. On such infrastructures, IPv6 services can be delivered by configuring IPv6 over ATM or frame relay permanent virtual circuits on attached IPv6 routers.

Looking ahead, the recent deployment of dense wavelength division multiplexing also could potentially facilitate the creation of an IPv6 topology by reserving wavelengths (lambdas) for IPv6 traffic.

IPv6 activities are numerous across many public and private organizations. An experimental infrastructure called the 6bone, created mostly by universities, network research labs and IP vendors, is now being tested and debugged using IPv6 protocols and operations (www.6bone.net). Presently it consists mostly of IPv6-over-IPv4 tunnels and encompasses more than 200 sites in 39 countries.

A production infrastructure in support of education and research called 6ren is being developed as well (www.6ren.net).

Making the Transition

Network managers who think the long-term time frame allows them to put off their IPv6 decisions should think again. It is not too early to begin planning, deploying and testing IPv6 networks. By preparing for the transition now instead of later, they can build a solid knowledge base and avoid the kind of frenzied response that characterized so many year-2000 programming efforts.

---

Shantz is Cisco Systems (www.cisco.com) Mobile Wireless Group vice president & general manager.

Cisco's involvement in IPv6 efforts includes chairing the IRTF IPNG Working Group, which is setting many of the IPv6 standards as well as the Ngtrans WG, which designed the transition tools between IPv4 and IPv6. Current status of these standards can be obtained from http://www.ietf.org/html.charters/ipngwg-charter.html. Cisco IOS IPv6 software also has been deployed in the prototype 6Bone network for testing purposes for the past five years.