IPv6

The Internet protocol (IP) address shortage crisis has been slowly materializing since the Internet became popular. A new numbering system, available in IP version 6 (IPv6), promises to help solve the problem, but faces significant challenges.

In the early 1990s, there were two challenges to the numbering system used to identify individual devices and networks. First, the IPv4 address space only supported 32 bits or 4.3 billion IP addresses. Second, not all IP addresses were created equal. Certain classes of IP addresses enabled an operator to create networks with far more devices than others, which led to bickering and competition for the more valuable ones. The creation of classless inter-domain routing (CIDR) in 1993 made all IP addresses more or less equal.

The next major step in helping to reduce the exhaustion of IP addresses was the advent of network address translation (NAT) in 1996. This essentially allowed a private network to reuse a single IP address across all of the devices behind the router.

While this practice reduces the number of IPv4 addresses needed, it also breaks the consistency of network routing. Hosts behind NAT routers cannot participate in some Internet protocols that require initiation of transmission control protocol (TCP) connections from the outside network or stateless protocols like user datagram protocol (UDP), which can be disrupted, unless the router has specific software to support the protocols. NAT can also complicate tunneling protocols for virtual private networks (VPNs).

Even with these advances, it appears that the number of unused IPv4 addresses is finally reaching exhaustion. Currently, only about 13 percent of the IPv4 address space remains. Estimates vary, but many predict we will run out of IPv4 addresses before 2012.

The adoption of IPv6 with a virtually limitless 128-bit address space – the number of available addresses works out to about 3.4 x 1038 or 34 followed by 37 zeroes – will solve the shortage. But the relief of the global IP address crisis comes with a complex blend of pain in enabling the new network to interoperate with the old.

In isolated networks where one operator controls all the variables, the move to IPv6 is straightforward. The challenge comes in getting the old network to talk to the new. Vendors have a number of solutions that address this problem that work well in some scenarios. But there are concerns that some of these patches could create problems down the road that will be difficult to troubleshoot or repair.

NTT’s virgin success

In virgin networks, IPv6 sounds like a dream come true, with easier network deployments and fewer challenges. A case in point is NTT, Japan’s incumbent telephone company and the third largest telecom in the world. An IPv6 early adopter¬ – its global NTT Com division went fully dual stack (IPv4/IPv6) in 2004¬¬ – NTT deployed an IPv6 fiber to the home (FTTH) network in March 2005, which now covers 89 percent of Japan.

There are currently more than 10 million data subscribers and hundreds of thousands of TV subscribers. The TV component, called Hikari-TV, provides 76 channels of standard definition (SD) and high definition (HD) TV, along with more than 10,000 video on demand (VOD) and 13,000 karaoke titles.

The use of IPv6 allows the network to support IP multicasting for the different TV channels. The distribution system sends one data stream for each channel to a neighborhood, which is multicast to individual homes. IPv4 can support multicast, but Cody Christman, director of product engineering for NTT America, said it is more complicated to set up because it does not provide end-to-end address transparency, and it has difficulty working across NAT.

The use of IPv6 also makes it easier to implement quality of service (QoS) so that throughput can be prioritized for each user and for applications like telephony, which require low latency. While IPv4 supports some QoS, Christman said that IPv6 provides enhancements like a flow label header that give providers more fined-grained control over how QoS features are implemented.

The only real challenge lies in providing interoperability to the outside Internet, via IPv4 tunneling. "It is different than the United States, where IPv4 is everywhere," Christman said. "In the United States, service providers are struggling to interoperate between IPv4 and IPv6 networks. This (the NTT system) is beautiful because it is a native IPv6 network, and a lot of the content is provided via IPv6 addresses."

Retrofit pain at Comcast

But the story for existing networks is not so sunny. In these cases, operators are faced with providing interoperability to customers’ existing IPv4 computers, gaming systems, printers and other devices, and to the existing IPv4 devices on the Internet at large.

Alain Durand, director of Internet governance and IPv6 architecture in the office of the CTO at Comcast, has spent years crafting the best transition strategy for the largest cable operator in the United States. The company now has 16.8 million digital customers with a total of 27 million digital set-top boxes, 14.7 million cable modems, and 6.1 million phone devices on its network.

In contrast with NTT’s view that IPv6 will enable all sorts of great new services and features, Comcast has a more minimalist outlook. "The only reason that IPv6 may be adopted is because we are running out of IPv4 addresses," Durand said.

"If there were enough IPv4 addresses, there would be no reason to go to IPv6. That is one of the reasons it has not been deployed more," he continued. "The No. 1 challenge is that the majority of traffic is IPv4. Changing the underlying technology is really hard. IPv6 offers more addresses and nothing else. It is not really a compelling proposal. The biggest technical challenge is that IPv4 and IPv6 are not compatible on the wire."

Vendors are starting to bring out more IPv6 products, Durand said. "One of the things the cable industry did was to recognize that IPv6 was going to be important, so a lot of work was done to bake it into new equipment standards like DOCSIS 3.0, PacketCable 2.0 and the OpenCable specifications for set-top boxes."

Comcast is exploring how to make the transition in stages. For example, it is starting out with using IPv6 to manage its network of cable modems. At this stage, only the management uses IPv6, while the Internet traffic is still IPv4. Management is the first to come out because it is easier and can operate in a contained environment.

The transition has helped Comcast understand all of the elements needed for a seamless transition.

"Comcast needed to upgrade the provision and monitoring system applications to communicate directly with the cable modems," Durand explained. "Back-office ‘glue’ with the billing systems is a complex part of the provisioning system. Until recently, these elements only supported IPv4. The biggest step Comcast has taken to make this work has been engaging the vendors to deliver the right pieces of software supporting IPv6."

When IPv6 first emerged, many assumed that the biggest deployment problems would be incorporating IPv6 at layer 3.

"Many believed that the hard problems with deploying IPv6 would be to deploy it on routers," Durand said. "It turned out this was much easier than expected. However, IPv6-aware applications (layer 7) were harder to find."

In the next phases, Comcast plans to enable IPv6 for Internet telephony and set-top boxes. This migration will take time.

"It is unclear in which orders the elements will come onto the network," Durand said. "We are trying to look at it in a holistic way. We will have some trials of all of those technologies in the years to come."

George Lawton is a free-lance writer.

Sidebar: CMTS and Modem Vendors Talk IPV6 Transition

From Steven Krapp, Director, CMTS Product Management, ARRIS Broadband:

"We support configuring interface addresses with IPv6, cable modems, including the eCM in both EMTAs and DSG set-top boxes, using IPv6 addresses, utilizing the DHCPv6 relay agent for the IPv6 cable modems, and static routing for the IPv6 subnets configured on the C4 CMTS. This is significant, as operators are suffering from IPv4 private address exhaustion. These are the addresses used to manage cable modems, EMTAs, and set-tops. With C4 CMTS release 7.0, it is possible for a cable operator to migrate the management of these devices from IPv4 private address space to IPv6. This potentially frees up millions of IPv4 private addresses, which the operator can then use elsewhere in their network."

From Ben Bekele, Marketing Manager, Cisco Systems:

"Cisco has been working on three fronts to enable and insure IPv6 readiness of its uBR7225VXR, uBR7246VXR, uBR10012 CMTS and the DPC3000 D3.0 CPEs to help partner MSOs enable IPv6 in their networks.

1. Service enablement and provisioning. Through DOCSIS 3.0 provisioning to support DHCPv6 Relay, IPv6 TFTP server, IPv6 ToD server, etc.; and enabled IPv6 and IPv4 provisioning, dual provisioning, DHCP relay features, alternative provisioning mode (APM).

2. Service/device management. Through support for Syslog over IPv6, SNMP over IPv6, MIBs updates for IPv6, Telnet over IPv6, SSH over IPv6, HTTP over IPv6, CLI enhancements for IPv6, etc.

3. Leveraging the aggregation and core routers IPv6 capability to validate IPv6 features and performance"

From Mike Cookish, Director of Product Management, Home and Network Mobility Solutions, Motorola:

"Motorola’s BSR 64000 is a fully redundant carrier-class intelligent edge router with an integrated DOCSIS 3.0 CMTS. The IPv6 implementation is CableLabs DOCSIS 3.0 Bronze Certified. As new IP-based services are delivered to subscribers, cable operators can migrate to DOCSIS 3.0 IPv6 to provide greater scalability and improved addressing via IPv6-capable cable modems and set-tops. Motorola’s Release 5.1 implementation supports simultaneous IPv4/IPv6 from the same device to support a smooth migration from IPv4 to IPv6 as some network management devices will not support IPv6 immediately. This enables cable modems to support both an IPv4 and IPv6 address initially."